1 systemd System and Service Manager
3 CHANGES WITH 230 in spe:
5 * DNSSEC is now turned on by default in systemd-resolved (in
6 "allow-downgrade" mode), but may be turned off during compile time by
7 passing "--with-default-dnssec=no" to "configure" (and of course,
8 during runtime with DNSSEC= in resolved.conf). We recommend
9 downstreams to leave this on at least during development cycles and
10 report any issues with the DNSSEC logic upstream. We are very
11 interested in collecting feedback about the DNSSEC validator and its
12 limitations in the wild. Note however, that DNSSEC support is
13 probably nothing downstreams should turn on in stable distros just
14 yet, as it might create incompatibilities with a few DNS servers and
15 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
16 automatically whenever we detect such incompatible setups, but there
17 might be systems we do not cover yet. Hence: please help us testing
18 the DNSSEC code, leave this on where you can, report back, but then
19 again don't consider turning this on in your stable, LTS or
20 production release just yet. (Note that you have to enable
21 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
22 and its DNSSEC mode for host name resolution from local
25 * systemd-resolve conveniently resolves DANE records with the --tlsa
26 option and OPENPGPKEY records with the --openpgp option. It also
27 supports dumping raw DNS record data via the new --raw= switch now.
29 * systemd-logind will now by default terminate user processes that are
30 part of the user session scope unit (session-XX.scope) when the user
31 logs out. This behavior is controlled by the KillUserProcesses=
32 setting in logind.conf, and the previous default of "no" is now
33 changed to "yes". This means that user sessions will be properly
34 cleaned up after, but additional steps are necessary to allow
35 intentionally long-running processes to survive logout.
37 While the user is logged in at least once, user@.service is running,
38 and any service that should survive the end of any individual login
39 session can be started at a user service or scope using systemd-run.
40 systemd-run(1) man page has been extended with an example which shows
41 how to run screen in a scope unit underneath user@.service. The same
42 command works for tmux.
44 After the user logs out of all sessions, user@.service will be
45 terminated too, by default, unless the user has "lingering" enabled.
46 To effectively allow users to run long-term tasks even if they are
47 logged out, lingering must be enabled for them. See loginctl(1) for
48 details. The default polkit policy was modified to allow users to
49 set lingering for themselves without authentication.
51 Previous defaults can be restored at compile time by the
52 --without-kill-user-processes option to "configure".
54 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
55 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
56 enable. Also, support for the "io" cgroup controller in the unified
57 hierarchy has been added, so that the "memory", "pids" and "io" are
58 now the controllers that are supported on the unified hierarchy.
60 WARNING: it is not possible to use previous systemd versions with
61 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
62 is necessary to also update systemd in the initramfs if using the
63 unified hierarchy. An updated SELinux policy is also required.
65 * LLDP support has been extended, and both passive (receive-only) and
66 active (sender) modes are supported. Passive mode ("routers-only") is
67 enabled by default in systemd-networkd. Active LLDP mode is enabled
68 by default for containers on the internal network. The "networkctl
69 lldp" command may be used to list information gathered. "networkctl
70 status" will also show basic LLDP information on connected peers now.
72 * The IAID and DUID unique identifier sent in DHCP requests may now be
73 configured for the system and each .network file managed by
76 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
77 systemd-socket-activate and installed into /usr/bin. It is now fully
80 * systemd-journald now uses separate threads to flush changes to disk
81 when closing journal files, thus reducing impact of slow disk I/O on
84 * systemd-ask-password now optionally skips printing of the password to
85 stdout with --no-output which can be useful in scripts.
87 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
88 (devices tagged with ID_MAKER_TOOL) are now tagged with
89 "uaccess" and are available to logged in users.
91 * systemd-bootchart has been split out to a separate repository:
92 https://github.com/systemd/systemd-bootchart
94 * The compatibility libraries libsystemd-daemon.so,
95 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
96 which have been deprecated since systemd-209 have been removed along
97 with the corresponding pkg-config files. All symbols provided by
98 those libraries are provided by libsystemd.so.
100 * The Capabilities= unit file setting has been removed (it is ignored
101 for backwards compatibility). AmbientCapabilities= and
102 CapabilityBoundingSet= should be used instead.
104 * "systemctl show" gained a new --value switch, which allows print a
105 only the contents of a specific unit property, without also printing
106 the property's name. Similar support was added to "show*" verbs
107 of loginctl and machinectl that output "key=value" lists.
109 * A new command "systemctl revert" has been added that may be used to
110 revert to the vendor version of a unit file, in case local changes
111 have been made by adding drop-ins or overriding the unit file.
113 * "machinectl clean" gained a new verb to automatically remove all or
114 just hidden container images.
116 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
117 merged into the kernel in its current form.
119 * systemd-networkd gained support for configuring proxy ARP support for
120 each interface, via the ProxyArp= setting in .network files. It also
121 gained support for configuring the multicast querier feature of
122 bridge devices, via the new MulticastQuerier= setting in .netdev
123 files. A new setting PreferredLifetime= has been added for addresses
124 configured in .network file to configure the lifetime intended for an
127 * systemd-tmpfiles gained support for a new line type "e" for emptying
128 directories, if they exist, without creating them if they don't.
130 * journalctl learned a new output mode "-o short-unix" that outputs log
131 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
132 UTC). It also gained support for a new --no-hostname setting to
133 suppress the hostname column in the family of "short" output modes.
135 * systemd-nspawn gained support for automatically patching the UID/GIDs
136 of the owners and the ACLs of all files and directories in a
137 container tree to match the UID/GID user namespacing range selected
138 for the container invocation. This mode is enabled via the new
139 --private-user-chown switch. It also gained support for automatically
140 choosing a free, previously unused UID/GID range when starting a
141 container, via the new --private-users=pick setting (which implies
142 --private-user-chown). Together, these options for the first time
143 make user namespacing for nspawn containers fully automatic and thus
144 deployable. The systemd-nspaw@.service template unit file has been
145 changed to use this functionality by default.
147 * systemd-nspawn gained a new --network-zone= switch, that allows
148 creating ad-hoc virtual Ethernet links between multiple containers,
149 that only exist as long as at least one container referencing them is
150 running. This allows easy connecting of multiple containers with a
151 common link that implements an Ethernet broadcast domain. Each of
152 these network "zones" may be named relatively freely by the user, and
153 may be referenced by any number of containers, but each container may
154 only reference one of these "zones". On the lower level, this is
155 implemented by an automatically managed bridge network interface for
156 each zone, that is created when the first container referencing its
157 zone is created and removed when the last one referencing its zone
160 * The default start timeout may now be configured on the kernel command
161 line via systemd.default_timeout_start_sec=. It was already
162 configurable via the DefaultTimeoutStartSec= option in
163 /etc/systemd/system.conf.
165 * Socket units gained a new TriggerLimitIntervalSec= and
166 TriggerLimitBurst= setting to configure a limit on the activation
167 rate of the socket unit.
169 * The LimitNICE= setting now optionally takes normal UNIX nice values
170 in addition to the raw integer limit value. If the specified
171 parameter is prefixed with "+" or "-" and is in the range -20..19 the
172 value is understood as UNIX nice value. If not prefixed like this it
173 is understood as raw RLIMIT_NICE limit.
175 * Note that the effect of the PrivateDevices= unit file setting changed
176 slightly with this release: the per-device /dev file system will be
177 mounted read-only from this version on, and will have "noexec"
178 set. This (minor) change of behaviour might cause some (exceptional)
179 legacy software to break, when PrivateDevices=yes is set for its
180 service. Please leave PrivateDevices= off if you run into problems
183 Contributions from: Alban Crequy, Alexander Kuleshov, Alex Crawford,
184 Andrew Eikum, Beniamino Galvani, Benjamin Robin, Benjamin ROBIN, Biao
185 Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Colin Guthrie, Daniel
186 J Walsh, Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
187 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
188 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
189 Bui, frankheckenbach, Georgia Brikis, Harald Hoyer, Hendrik Brueckner,
190 Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo Puustinen, Jakub
191 Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth, kayrus, Klearchos
192 Chaloulos, Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukáš
193 Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt, Michael Biebl,
194 michaelolbrich, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletar,
195 Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin, mulkieran,
196 muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween, Nicolas
197 Braud-Santoni, Patrik Flykt, Peter Hutterer, Petr Lautrbach, Petros
198 Angelatos, Piotr Drąg, Rabin Vincent, Robert Węcławski, Ronny
199 Chevalier, Samuel Tardieu, Stefan Schallenberg, Steven Siloti, Susant
200 Sahani, Sylvain Plantefève, Taylor Smock, tblume, Tejun Heo, Thomas
201 Blume, Thomas Haller, Thomas Hindoe Paaboel Andersen, Thomas
202 H. P. Andersen, Tobias Klauser, Tom Gundersen, Torstein Husebø, Umut
203 Tezduyar Lindskog, Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam),
204 Vladimir Panteleev, Wieland Hoffmann, Wouter Verhelst, Yu Watanabe,
205 Zbigniew Jędrzejewski-Szmek
211 * The systemd-resolved DNS resolver service has gained a substantial
212 set of new features, most prominently it may now act as a DNSSEC
213 validating stub resolver. DNSSEC mode is currently turned off by
214 default, but is expected to be turned on by default in one of the
215 next releases. For now, we invite everybody to test the DNSSEC logic
216 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
217 service also gained a full set of D-Bus interfaces, including calls
218 to configure DNS and DNSSEC settings per link (for use by external
219 network management software). systemd-resolved and systemd-networkd
220 now distinguish between "search" and "routing" domains. The former
221 are used to qualify single-label names, the latter are used purely
222 for routing lookups within certain domains to specific links.
223 resolved now also synthesizes RRs for all entries from /etc/hosts.
225 * The systemd-resolve tool (which is a client utility for
226 systemd-resolved) has been improved considerably and is now fully
227 supported and documented. Hence it has moved from /usr/lib/systemd to
230 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
233 * The coredump collection logic has been reworked: when a coredump is
234 collected it is now written to disk, compressed and processed
235 (including stacktrace extraction) from a new instantiated service
236 systemd-coredump@.service, instead of directly from the
237 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
238 processing large coredumps can take up a substantial amount of
239 resources and time, and this previously happened entirely outside of
240 systemd's service supervision. With the new logic the core_pattern
241 hook only does minimal metadata collection before passing off control
242 to the new instantiated service, which is configured with a time
243 limit, a nice level and other settings to minimize negative impact on
244 the rest of the system. Also note that the new logic will honour the
245 RLIMIT_CORE setting of the crashed process, which now allows users
246 and processes to turn off coredumping for their processes by setting
249 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
250 and all forked processes by default. Previously, PID 1 would leave
251 the setting at "0" for all processes, as set by the kernel. Note that
252 the resource limit traditionally has no effect on the generated
253 coredumps on the system if the /proc/sys/kernel/core_pattern hook
254 logic is used. Since the limit is now honoured (see above) its
255 default has been changed so that the coredumping logic is enabled by
256 default for all processes, while allowing specific opt-out.
258 * When the stacktrace is extracted from processes of system users, this
259 is now done as "systemd-coredump" user, in order to sandbox this
260 potentially security sensitive parsing operation. (Note that when
261 processing coredumps of normal users this is done under the user ID
262 of process that crashed, as before.) Packagers should take notice
263 that it is now necessary to create the "systemd-coredump" system user
264 and group at package installation time.
266 * The systemd-activate socket activation testing tool gained support
267 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
268 and --seqpacket switches. It also has been extended to support both
269 new-style and inetd-style file descriptor passing. Use the new
270 --inetd switch to request inetd-style file descriptor passing.
272 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
273 variable, which takes a boolean value. If set to false, ANSI color
274 output is disabled in the tools even when run on a terminal that
277 * The VXLAN support in networkd now supports two new settings
278 DestinationPort= and PortRange=.
280 * A new systemd.machine_id= kernel command line switch has been added,
281 that may be used to set the machine ID in /etc/machine-id if it is
282 not initialized yet. This command line option has no effect if the
283 file is already initialized.
285 * systemd-nspawn gained a new --as-pid2 switch that invokes any
286 specified command line as PID 2 rather than PID 1 in the
287 container. In this mode PID 1 is a minimal stub init process that
288 implements the special POSIX and Linux semantics of PID 1 regarding
289 signal and child process management. Note that this stub init process
290 is implemented in nspawn itself and requires no support from the
291 container image. This new logic is useful to support running
292 arbitrary commands in the container, as normal processes are
293 generally not prepared to run as PID 1.
295 * systemd-nspawn gained a new --chdir= switch for setting the current
296 working directory for the process started in the container.
298 * "journalctl /dev/sda" will now output all kernel log messages for
299 specified device from the current boot, in addition to all devices
300 that are parents of it. This should make log output about devices
301 pretty useful, as long as kernel drivers attach enough metadata to
302 the log messages. (The usual SATA drivers do.)
304 * The sd-journal API gained two new calls
305 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
306 that report whether log data from /run or /var has been found.
308 * journalctl gained a new switch "--fields" that prints all journal
309 record field names currently in use in the journal. This is backed
310 by two new sd-journal API calls sd_journal_enumerate_fields() and
311 sd_journal_restart_fields().
313 * Most configurable timeouts in systemd now expect an argument of
314 "infinity" to turn them off, instead of "0" as before. The semantics
315 from now on is that a timeout of "0" means "now", and "infinity"
316 means "never". To maintain backwards compatibility, "0" continues to
317 turn off previously existing timeout settings.
319 * "systemctl reload-or-try-restart" has been renamed to "systemctl
320 try-reload-or-restart" to clarify what it actually does: the "try"
321 logic applies to both reloading and restarting, not just restarting.
322 The old name continues to be accepted for compatibility.
324 * On boot-up, when PID 1 detects that the system clock is behind the
325 release date of the systemd version in use, the clock is now set
326 to the latter. Previously, this was already done in timesyncd, in order
327 to avoid running with clocks set to the various clock epochs such as
328 1902, 1938 or 1970. With this change the logic is now done in PID 1
329 in addition to timesyncd during early boot-up, so that it is enforced
330 before the first process is spawned by systemd. Note that the logic
331 in timesyncd remains, as it is more comprehensive and ensures
332 clock monotonicity by maintaining a persistent timestamp file in
333 /var. Since /var is generally not available in earliest boot or the
334 initrd, this part of the logic remains in timesyncd, and is not done
337 * Support for tweaking details in net_cls.class_id through the
338 NetClass= configuration directive has been removed, as the kernel
339 people have decided to deprecate that controller in cgroup v2.
340 Userspace tools such as nftables are moving over to setting rules
341 that are specific to the full cgroup path of a task, which obsoletes
342 these controllers anyway. The NetClass= directive is kept around for
343 legacy compatibility reasons. For a more in-depth description of the
344 kernel change, please refer to the respective upstream commit:
346 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
348 * A new service setting RuntimeMaxSec= has been added that may be used
349 to specify a maximum runtime for a service. If the timeout is hit, the
350 service is terminated and put into a failure state.
352 * A new service setting AmbientCapabilities= has been added. It allows
353 configuration of additional Linux process capabilities that are
354 passed to the activated processes. This is only available on very
357 * The process resource limit settings in service units may now be used
358 to configure hard and soft limits individually.
360 * The various libsystemd APIs such as sd-bus or sd-event now publicly
361 expose support for gcc's __attribute__((cleanup())) C extension.
362 Specifically, for many object destructor functions alternative
363 versions have been added that have names suffixed with "p" and take a
364 pointer to a pointer to the object to destroy, instead of just a
365 pointer to the object itself. This is useful because these destructor
366 functions may be used directly as parameters to the cleanup
367 construct. Internally, systemd has been a heavy user of this GCC
368 extension for a long time, and with this change similar support is
369 now available to consumers of the library outside of systemd. Note
370 that by using this extension in your sources compatibility with old
371 and strictly ANSI compatible C compilers is lost. However, all gcc or
372 LLVM versions of recent years support this extension.
374 * Timer units gained support for a new setting RandomizedDelaySec= that
375 allows configuring some additional randomized delay to the configured
376 time. This is useful to spread out timer events to avoid load peaks in
377 clusters or larger setups.
379 * Calendar time specifications now support sub-second accuracy.
381 * Socket units now support listening on SCTP and UDP-lite protocol
384 * The sd-event API now comes with a full set of man pages.
386 * Older versions of systemd contained experimental support for
387 compressing journal files and coredumps with the LZ4 compressor that
388 was not compatible with the lz4 binary (due to API limitations of the
389 lz4 library). This support has been removed; only support for files
390 compatible with the lz4 binary remains. This LZ4 logic is now
391 officially supported and no longer considered experimental.
393 * The dkr image import logic has been removed again from importd. dkr's
394 micro-services focus doesn't fit into the machine image focus of
395 importd, and quickly got out of date with the upstream dkr API.
397 * Creation of the /run/lock/lockdev/ directory was dropped from
398 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
399 been available for many years. If you still need this, you need to
400 create your own tmpfiles.d config file with:
402 d /run/lock/lockdev 0775 root lock -
404 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
405 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
406 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
407 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
408 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
409 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
410 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
411 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
412 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
413 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
414 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
415 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
416 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
417 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
418 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
419 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
420 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
421 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
422 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
428 * A number of properties previously only settable in unit
429 files are now also available as properties to set when
430 creating transient units programmatically via the bus, as it
431 is exposed with systemd-run's --property=
432 setting. Specifically, these are: SyslogIdentifier=,
433 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
434 EnvironmentFile=, ReadWriteDirectories=,
435 ReadOnlyDirectories=, InaccessibleDirectories=,
436 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
438 * When creating transient services via the bus API it is now
439 possible to pass in a set of file descriptors to use as
440 STDIN/STDOUT/STDERR for the invoked process.
442 * Slice units may now be created transiently via the bus APIs,
443 similar to the way service and scope units may already be
446 * Wherever systemd expects a calendar timestamp specification
447 (like in journalctl's --since= and --until= switches) UTC
448 timestamps are now supported. Timestamps suffixed with "UTC"
449 are now considered to be in Universal Time Coordinated
450 instead of the local timezone. Also, timestamps may now
451 optionally be specified with sub-second accuracy. Both of
452 these additions also apply to recurring calendar event
453 specification, such as OnCalendar= in timer units.
455 * journalctl gained a new "--sync" switch that asks the
456 journal daemon to write all so far unwritten log messages to
457 disk and sync the files, before returning.
459 * systemd-tmpfiles learned two new line types "q" and "Q" that
460 operate like "v", but also set up a basic btrfs quota
461 hierarchy when used on a btrfs file system with quota
464 * tmpfiles' "v", "q" and "Q" will now create a plain directory
465 instead of a subvolume (even on a btrfs file system) if the
466 root directory is a plain directory, and not a
467 subvolume. This should simplify things with certain chroot()
468 environments which are not aware of the concept of btrfs
471 * systemd-detect-virt gained a new --chroot switch to detect
472 whether execution takes place in a chroot() environment.
474 * CPUAffinity= now takes CPU index ranges in addition to
477 * The various memory-related resource limit settings (such as
478 LimitAS=) now understand the usual K, M, G, ... suffixes to
479 the base of 1024 (IEC). Similar, the time-related resource
480 limit settings understand the usual min, h, day, ...
483 * There's a new system.conf setting DefaultTasksMax= to
484 control the default TasksMax= setting for services and
485 scopes running on the system. (TasksMax= is the primary
486 setting that exposes the "pids" cgroup controller on systemd
487 and was introduced in the previous systemd release.) The
488 setting now defaults to 512, which means services that are
489 not explicitly configured otherwise will only be able to
490 create 512 processes or threads at maximum, from this
491 version on. Note that this means that thread- or
492 process-heavy services might need to be reconfigured to set
493 TasksMax= to a higher value. It is sufficient to set
494 TasksMax= in these specific unit files to a higher value, or
495 even "infinity". Similar, there's now a logind.conf setting
496 UserTasksMax= that defaults to 4096 and limits the total
497 number of processes or tasks each user may own
498 concurrently. nspawn containers also have the TasksMax=
499 value set by default now, to 8192. Note that all of this
500 only has an effect if the "pids" cgroup controller is
501 enabled in the kernel. The general benefit of these changes
502 should be a more robust and safer system, that provides a
503 certain amount of per-service fork() bomb protection.
505 * systemd-nspawn gained the new --network-veth-extra= switch
506 to define additional and arbitrarily-named virtual Ethernet
507 links between the host and the container.
509 * A new service execution setting PassEnvironment= has been
510 added that allows importing select environment variables
511 from PID1's environment block into the environment block of
514 * Timer units gained support for a new RemainAfterElapse=
515 setting which takes a boolean argument. It defaults on on,
516 exposing behaviour unchanged to previous releases. If set to
517 off, timer units are unloaded after they elapsed if they
518 cannot elapse again. This is particularly useful for
519 transient timer units, which shall not stay around longer
520 than until they first elapse.
522 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
523 default now (the kernel default is 16). This is beneficial
524 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
525 allows substantially larger numbers of queued
526 datagrams. This should increase the capability of systemd to
527 parallelize boot-up, as logging and sd_notify() are unlikely
528 to stall execution anymore. If you need to change the value
529 from the new defaults, use the usual sysctl.d/ snippets.
531 * The compression framing format used by the journal or
532 coredump processing has changed to be in line with what the
533 official LZ4 tools generate. LZ4 compression support in
534 systemd was considered unsupported previously, as the format
535 was not compatible with the normal tools. With this release
536 this has changed now, and it is hence safe for downstream
537 distributions to turn it on. While not compressing as well
538 as the XZ, LZ4 is substantially faster, which makes
539 it a good default choice for the compression logic in the
540 journal and in coredump handling.
542 * Any reference to /etc/mtab has been dropped from
543 systemd. The file has been obsolete since a while, but
544 systemd refused to work on systems where it was incorrectly
545 set up (it should be a symlink or non-existent). Please make
546 sure to update to util-linux 2.27.1 or newer in conjunction
547 with this systemd release, which also drops any reference to
548 /etc/mtab. If you maintain a distribution make sure that no
549 software you package still references it, as this is a
550 likely source of bugs. There's also a glibc bug pending,
551 asking for removal of any reference to this obsolete file:
553 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
555 Note that only util-linux versions built with
556 --enable-libmount-force-mountinfo are supported.
558 * Support for the ".snapshot" unit type has been removed. This
559 feature turned out to be little useful and little used, and
560 has now been removed from the core and from systemctl.
562 * The dependency types RequiresOverridable= and
563 RequisiteOverridable= have been removed from systemd. They
564 have been used only very sparingly to our knowledge and
565 other options that provide a similar effect (such as
566 systemctl --mode=ignore-dependencies) are much more useful
567 and commonly used. Moreover, they were only half-way
568 implemented as the option to control behaviour regarding
569 these dependencies was never added to systemctl. By removing
570 these dependency types the execution engine becomes a bit
571 simpler. Unit files that use these dependencies should be
572 changed to use the non-Overridable dependency types
573 instead. In fact, when parsing unit files with these
574 options, that's what systemd will automatically convert them
575 too, but it will also warn, asking users to fix the unit
576 files accordingly. Removal of these dependency types should
577 only affect a negligible number of unit files in the wild.
579 * Behaviour of networkd's IPForward= option changed
580 (again). It will no longer maintain a per-interface setting,
581 but propagate one way from interfaces where this is enabled
582 to the global kernel setting. The global setting will be
583 enabled when requested by a network that is set up, but
584 never be disabled again. This change was made to make sure
585 IPv4 and IPv6 behaviour regarding packet forwarding is
586 similar (as the Linux IPv6 stack does not support
587 per-interface control of this setting) and to minimize
590 * In unit files the behaviour of %u, %U, %h, %s has
591 changed. These specifiers will now unconditionally resolve
592 to the various user database fields of the user that the
593 systemd instance is running as, instead of the user
594 configured in the specific unit via User=. Note that this
595 effectively doesn't change much, as resolving of these
596 specifiers was already turned off in the --system instance
597 of systemd, as we cannot do NSS lookups from PID 1. In the
598 --user instance of systemd these specifiers where correctly
599 resolved, but hardly made any sense, since the user instance
600 lacks privileges to do user switches anyway, and User= is
601 hence useless. Morever, even in the --user instance of
602 systemd behaviour was awkward as it would only take settings
603 from User= assignment placed before the specifier into
604 account. In order to unify and simplify the logic around
605 this the specifiers will now always resolve to the
606 credentials of the user invoking the manager (which in case
607 of PID 1 is the root user).
609 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
610 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
611 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
612 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
613 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
614 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
615 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
616 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
617 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
618 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
619 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
620 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
621 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
622 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
629 * systemd now depends on util-linux v2.27. More specifically,
630 the newly added mount monitor feature in libmount now
631 replaces systemd's former own implementation.
633 * libmount mandates /etc/mtab not to be regular file, and
634 systemd now enforces this condition at early boot.
635 /etc/mtab has been deprecated and warned about for a very
636 long time, so systems running systemd should already have
637 stopped having this file around as anything else than a
638 symlink to /proc/self/mounts.
640 * Support for the "pids" cgroup controller has been added. It
641 allows accounting the number of tasks in a cgroup and
642 enforcing limits on it. This adds two new setting
643 TasksAccounting= and TasksMax= to each unit, as well as a
644 global option DefaultTasksAccounting=.
646 * Support for the "net_cls" cgroup controller has been added.
647 It allows assigning a net class ID to each task in the
648 cgroup, which can then be used in firewall rules and traffic
649 shaping configurations. Note that the kernel netfilter net
650 class code does not currently work reliably for ingress
651 packets on unestablished sockets.
653 This adds a new config directive called NetClass= to CGroup
654 enabled units. Allowed values are positive numbers for fixed
655 assignments and "auto" for picking a free value
658 * 'systemctl is-system-running' now returns 'offline' if the
659 system is not booted with systemd. This command can now be
660 used as a substitute for 'systemd-notify --booted'.
662 * Watchdog timeouts have been increased to 3 minutes for all
663 in-tree service files. Apparently, disk IO issues are more
664 frequent than we hoped, and user reported >1 minute waiting
667 * 'machine-id-commit' functionality has been merged into
668 'machine-id-setup --commit'. The separate binary has been
671 * The WorkingDirectory= directive in unit files may now be set
672 to the special value '~'. In this case, the working
673 directory is set to the home directory of the user
676 * "machinectl shell" will now open the shell in the home
677 directory of the selected user by default.
679 * The CrashChVT= configuration file setting is renamed to
680 CrashChangeVT=, following our usual logic of not
681 abbreviating unnecessarily. The old directive is still
682 supported for compat reasons. Also, this directive now takes
683 an integer value between 1 and 63, or a boolean value. The
684 formerly supported '-1' value for disabling stays around for
687 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
688 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
689 RootDirectory= properties can now be set for transient
692 * The systemd-analyze tool gained a new "set-log-target" verb
693 to change the logging target the system manager logs to
694 dynamically during runtime. This is similar to how
695 "systemd-analyze set-log-level" already changes the log
698 * In nspawn /sys is now mounted as tmpfs, with only a selected
699 set of subdirectories mounted in from the real sysfs. This
700 enhances security slightly, and is useful for ensuring user
701 namespaces work correctly.
703 * Support for USB FunctionFS activation has been added. This
704 allows implementation of USB gadget services that are
705 activated as soon as they are requested, so that they don't
706 have to run continously, similar to classic socket
709 * The "systemctl exit" command now optionally takes an
710 additional parameter that sets the exit code to return from
711 the systemd manager when exiting. This is only relevant when
712 running the systemd user instance, or when running the
713 system instance in a container.
715 * sd-bus gained the new API calls sd_bus_path_encode_many()
716 and sd_bus_path_decode_many() that allow easy encoding and
717 decoding of multiple identifier strings inside a D-Bus
718 object path. Another new call sd_bus_default_flush_close()
719 has been added to flush and close per-thread default
722 * systemd-cgtop gained support for a -M/--machine= switch to
723 show the control groups within a certain container only.
725 * "systemctl kill" gained support for an optional --fail
726 switch. If specified the requested operation will fail of no
727 processes have been killed, because the unit had no
728 processes attached, or similar.
730 * A new systemd.crash_reboot=1 kernel command line option has
731 been added that triggers a reboot after crashing. This can
732 also be set through CrashReboot= in systemd.conf.
734 * The RuntimeDirectory= setting now understands unit
735 specifiers like %i or %f.
737 * A new (still internal) libary API sd-ipv4acd has been added,
738 that implements address conflict detection for IPv4. It's
739 based on code from sd-ipv4ll, and will be useful for
740 detecting DHCP address conflicts.
742 * File descriptors passed during socket activation may now be
743 named. A new API sd_listen_fds_with_names() is added to
744 access the names. The default names may be overridden,
745 either in the .socket file using the FileDescriptorName=
746 parameter, or by passing FDNAME= when storing the file
747 descriptors using sd_notify().
749 * systemd-networkd gained support for:
751 - Setting the IPv6 Router Advertisment settings via
752 IPv6AcceptRouterAdvertisements= in .network files.
754 - Configuring the HelloTimeSec=, MaxAgeSec= and
755 ForwardDelaySec= bridge parameters in .netdev files.
757 - Configuring PreferredSource= for static routes in
760 * The "ask-password" framework used to query for LUKS harddisk
761 passwords or SSL passwords during boot gained support for
762 caching passwords in the kernel keyring, if it is
763 available. This makes sure that the user only has to type in
764 a passphrase once if there are multiple objects to unlock
765 with the same one. Previously, such password caching was
766 available only when Plymouth was used; this moves the
767 caching logic into the systemd codebase itself. The
768 "systemd-ask-password" utility gained a new --keyname=
769 switch to control which kernel keyring key to use for
770 caching a password in. This functionality is also useful for
771 enabling display managers such as gdm to automatically
772 unlock the user's GNOME keyring if its passphrase, the
773 user's password and the harddisk password are the same, if
774 gdm-autologin is used.
776 * When downloading tar or raw images using "machinectl
777 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
778 file is now also downloaded, if it is available and stored
779 next to the image file.
781 * Units of type ".socket" gained a new boolean setting
782 Writable= which is only useful in conjunction with
783 ListenSpecial=. If true, enables opening the specified
784 special file in O_RDWR mode rather than O_RDONLY mode.
786 * systemd-rfkill has been reworked to become a singleton
787 service that is activated through /dev/rfkill on each rfkill
788 state change and saves the settings to disk. This way,
789 systemd-rfkill is now compatible with devices that exist
790 only intermittendly, and even restores state if the previous
791 system shutdown was abrupt rather than clean.
793 * The journal daemon gained support for vacuuming old journal
794 files controlled by the number of files that shall remain,
795 in addition to the already existing control by size and by
796 date. This is useful as journal interleaving performance
797 degrades with too many seperate journal files, and allows
798 putting an effective limit on them. The new setting defaults
799 to 100, but this may be changed by setting SystemMaxFiles=
800 and RuntimeMaxFiles= in journald.conf. Also, the
801 "journalctl" tool gained the new --vacuum-files= switch to
802 manually vacuum journal files to leave only the specified
803 number of files in place.
805 * udev will now create /dev/disk/by-path links for ATA devices
806 on kernels where that is supported.
808 * Galician, Serbian, Turkish and Korean translations were added.
810 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
811 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
812 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
813 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
814 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
815 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
816 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
817 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
818 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
819 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
820 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
821 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
822 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
823 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
824 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
825 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
826 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
827 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
833 * The DHCP implementation of systemd-networkd gained a set of
836 - The DHCP server now supports emitting DNS and NTP
837 information. It may be enabled and configured via
838 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
839 and NTP information is enabled, but no servers are
840 configured, the corresponding uplink information (if there
841 is any) is propagated.
843 - Server and client now support transmission and reception
844 of timezone information. It can be configured via the
845 newly introduced network options UseTimezone=,
846 EmitTimezone=, and Timezone=. Transmission of timezone
847 information is enabled between host and containers by
848 default now: the container will change its local timezone
849 to what the host has set.
851 - Lease timeouts can now be configured via
852 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
854 - The DHCP server improved on the stability of
855 leases. Clients are more likely to get the same lease
856 information back, even if the server loses state.
858 - The DHCP server supports two new configuration options to
859 control the lease address pool metrics, PoolOffset= and
862 * The encapsulation limit of tunnels in systemd-networkd may
863 now be configured via 'EncapsulationLimit='. It allows
864 modifying the maximum additional levels of encapsulation
865 that are permitted to be prepended to a packet.
867 * systemd now supports the concept of user buses replacing
868 session buses, if used with dbus-1.10 (and enabled via dbus
869 --enable-user-session). It previously only supported this on
870 kdbus-enabled systems, and this release expands this to
871 'dbus-daemon' systems.
873 * systemd-networkd now supports predictable interface names
876 * systemd now optionally supports the new Linux kernel
877 "unified" control group hierarchy. If enabled via the kernel
878 command-line option 'systemd.unified_cgroup_hierarchy=1',
879 systemd will try to mount the unified cgroup hierarchy
880 directly on /sys/fs/cgroup. If not enabled, or not
881 available, systemd will fall back to the legacy cgroup
882 hierarchy setup, as before. Host system and containers can
883 mix and match legacy and unified hierarchies as they
884 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
885 environment variable to individually select the hierarchy to
886 use for executed containers. By default, nspawn will use the
887 unified hierarchy for the containers if the host uses the
888 unified hierarchy, and the legacy hierarchy otherwise.
889 Please note that at this point the unified hierarchy is an
890 experimental kernel feature and is likely to change in one
891 of the next kernel releases. Therefore, it should not be
892 enabled by default in downstream distributions yet. The
893 minimum required kernel version for the unified hierarchy to
894 work is 4.2. Note that when the unified hierarchy is used
895 for the first time delegated access to controllers is
896 safe. Because of this systemd-nspawn containers will get
897 access to controllers now, as will systemd user
898 sessions. This means containers and user sessions may now
899 manage their own resources, partitioning up what the system
902 * A new special scope unit "init.scope" has been introduced
903 that encapsulates PID 1 of the system. It may be used to
904 determine resource usage and enforce resource limits on PID
905 1 itself. PID 1 hence moved out of the root of the control
908 * The cgtop tool gained support for filtering out kernel
909 threads when counting tasks in a control group. Also, the
910 count of processes is now recursively summed up by
911 default. Two options -k and --recursive= have been added to
912 revert to old behaviour. The tool has also been updated to
913 work correctly in containers now.
915 * systemd-nspawn's --bind= and --bind-ro= options have been
916 extended to allow creation of non-recursive bind mounts.
918 * libsystemd gained two new calls sd_pid_get_cgroup() and
919 sd_peer_get_cgroup() which return the control group path of
920 a process or peer of a connected AF_UNIX socket. This
921 function call is particularly useful when implementing
922 delegated subtrees support in the control group hierarchy.
924 * The "sd-event" event loop API of libsystemd now supports
925 correct dequeuing of real-time signals, without losing
928 * When systemd requests a PolicyKit decision when managing
929 units it will now add additional fields to the request,
930 including unit name and desired operation. This enables more
931 powerful PolicyKit policies, that make decisions depending
934 * nspawn learnt support for .nspawn settings files, that may
935 accompany the image files or directories of containers, and
936 may contain additional settings for the container. This is
937 an alternative to configuring container parameters via the
940 Contributions from: Cristian Rodríguez, Daniel Mack, David
941 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
942 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
943 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
944 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
945 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
946 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
947 Andersen, Tom Gundersen, Torstein Husebø
953 * machinectl gained a new verb 'shell' which opens a fresh
954 shell on the target container or the host. It is similar to
955 the existing 'login' command of machinectl, but spawns the
956 shell directly without prompting for username or
957 password. The pseudo machine '.host' now refers to the local
958 host and is used by default. Hence, 'machinectl shell' can
959 be used as replacement for 'su -' which spawns a session as
960 a fresh systemd unit in a way that is fully isolated from
961 the originating session.
963 * systemd-networkd learned to cope with private-zone DHCP
964 options and allows other programs to query the values.
966 * SELinux access control when enabling/disabling units is no
967 longer enforced with this release. The previous
968 implementation was incorrect, and a new corrected
969 implementation is not yet available. As unit file operations
970 are still protected via PolicyKit and D-Bus policy this is
971 not a security problem. Yet, distributions which care about
972 optimal SELinux support should probably not stabilize on
975 * sd-bus gained support for matches of type "arg0has=", that
976 test for membership of strings in string arrays sent in bus
979 * systemd-resolved now dumps the contents of its DNS and LLMNR
980 caches to the logs on reception of the SIGUSR1 signal. This
981 is useful to debug DNS behaviour.
983 * The coredumpctl tool gained a new --directory= option to
984 operate on journal files in a specific directory.
986 * "systemctl reboot" and related commands gained a new
987 "--message=" option which may be used to set a free-text
988 wall message when shutting down or rebooting the
989 system. This message is also logged, which is useful for
990 figuring out the reason for a reboot or shutdown a
993 * The "systemd-resolve-host" tool's -i switch now takes
994 network interface numbers as alternative to interface names.
996 * A new unit file setting for services has been introduced:
997 UtmpMode= allows configuration of how precisely systemd
998 handles utmp and wtmp entries for the service if this is
999 enabled. This allows writing services that appear similar to
1000 user sessions in the output of the "w", "who", "last" and
1003 * systemd-resolved will now locally synthesize DNS resource
1004 records for the "localhost" and "gateway" domains as well as
1005 the local hostname. This should ensure that clients querying
1006 RRs via resolved will get similar results as those going via
1007 NSS, if nss-myhostname is enabled.
1009 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1010 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1011 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1012 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1013 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1014 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1015 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1016 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1017 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1018 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1019 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1020 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1022 — Berlin, 2015-08-27
1026 * The systemd-efi-boot-generator functionality was merged into
1027 systemd-gpt-auto-generator.
1029 * systemd-networkd now supports Group Policy for vxlan
1030 devices. It can be enabled via the new boolean configuration
1031 option called 'GroupPolicyExtension='.
1033 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1034 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1035 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1037 — Berlin, 2015-07-31
1041 * The python-systemd code has been removed from the systemd repository.
1042 A new repository has been created which accommodates the code from
1043 now on, and we kindly ask distributions to create a separate package
1044 for this: https://github.com/systemd/python-systemd
1046 * The systemd daemon will now reload its main configuration
1047 (/etc/systemd/system.conf) on daemon-reload.
1049 * sd-dhcp now exposes vendor specific extensions via
1050 sd_dhcp_lease_get_vendor_specific().
1052 * systemd-networkd gained a number of new configuration options.
1054 - A new boolean configuration option for TAP devices called
1055 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1056 device, thus allowing to send and receive GSO packets.
1058 - A new tunnel configuration option called 'CopyDSCP='.
1059 If enabled, the DSCP field of ip6 tunnels is copied into the
1060 decapsulated packet.
1062 - A set of boolean bridge configuration options were added.
1063 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1064 and 'UnicastFlood=' are now parsed by networkd and applied to the
1065 respective bridge link device via the respective IFLA_BRPORT_*
1068 - A new string configuration option to override the hostname sent
1069 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1070 is true, networkd will use the configured hostname instead of the
1071 system hostname when sending DHCP requests.
1073 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1074 networkd will configure the IPv6 flow-label of the tunnel device
1075 according to RFC2460.
1077 - The 'macvtap' virtual network devices are now supported, similar to
1078 the already supported 'macvlan' devices.
1080 * systemd-resolved now implements RFC5452 to improve resilience against
1081 cache poisoning. Additionally, source port randomization is enabled
1082 by default to further protect against DNS spoofing attacks.
1084 * nss-mymachines now supports translating UIDs and GIDs of running
1085 containers with user-namespaces enabled. If a container 'foo'
1086 translates a host uid 'UID' to the container uid 'TUID', then
1087 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1088 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1089 mapped as 'vg-foo-TGID'.
1091 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1092 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1093 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1094 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1095 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1096 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1097 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1098 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1099 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1100 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1102 — Berlin, 2015-07-29
1106 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1107 There are no known issues with current sysfs, and udev does not need
1108 or should be used to work around such bugs.
1110 * udev does no longer enable USB HID power management. Several reports
1111 indicate, that some devices cannot handle that setting.
1113 * The udev accelerometer helper was removed. The functionality
1114 is now fully included in iio-sensor-proxy. But this means,
1115 older iio-sensor-proxy versions will no longer provide
1116 accelerometer/orientation data with this systemd version.
1117 Please upgrade iio-sensor-proxy to version 1.0.
1119 * networkd gained a new configuration option IPv6PrivacyExtensions=
1120 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1121 for Stateless Address") on selected networks.
1123 * For the sake of fewer build-time dependencies and less code in the
1124 main repository, the python bindings are about to be removed in the
1125 next release. A new repository has been created which accommodates
1126 the code from now on, and we kindly ask distributions to create a
1127 separate package for this. The removal will take place in v223.
1129 https://github.com/systemd/python-systemd
1131 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1132 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1133 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1134 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1135 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1136 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1137 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1138 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1139 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1140 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1142 — Berlin, 2015-07-07
1146 * The sd-bus.h and sd-event.h APIs have now been declared
1147 stable and have been added to the official interface of
1148 libsystemd.so. sd-bus implements an alternative D-Bus client
1149 library, that is relatively easy to use, very efficient and
1150 supports both classic D-Bus as well as kdbus as transport
1151 backend. sd-event is a generic event loop abstraction that
1152 is built around Linux epoll, but adds features such as event
1153 prioritization or efficient timer handling. Both APIs are good
1154 choices for C programs looking for a bus and/or event loop
1155 implementation that is minimal and does not have to be
1156 portable to other kernels.
1158 * kdbus support is no longer compile-time optional. It is now
1159 always built-in. However, it can still be disabled at
1160 runtime using the kdbus=0 kernel command line setting, and
1161 that setting may be changed to default to off, by specifying
1162 --disable-kdbus at build-time. Note though that the kernel
1163 command line setting has no effect if the kdbus.ko kernel
1164 module is not installed, in which case kdbus is (obviously)
1165 also disabled. We encourage all downstream distributions to
1166 begin testing kdbus by adding it to the kernel images in the
1167 development distributions, and leaving kdbus support in
1170 * The minimal required util-linux version has been bumped to
1173 * Support for chkconfig (--enable-chkconfig) was removed in
1174 favor of calling an abstraction tool
1175 /lib/systemd/systemd-sysv-install. This needs to be
1176 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1177 in README for details.
1179 * If there's a systemd unit and a SysV init script for the
1180 same service name, and the user executes "systemctl enable"
1181 for it (or a related call), then this will now enable both
1182 (or execute the related operation on both), not just the
1185 * The libudev API documentation has been converted from gtkdoc
1188 * gudev has been removed from the systemd tree, it is now an
1191 * The systemd-cgtop tool learnt a new --raw switch to generate
1192 "raw" (machine parsable) output.
1194 * networkd's IPForwarding= .network file setting learnt the
1195 new setting "kernel", which ensures that networkd does not
1196 change the IP forwarding sysctl from the default kernel
1199 * The systemd-logind bus API now exposes a new boolean
1200 property "Docked" that reports whether logind considers the
1201 system "docked", i.e. connected to a docking station or not.
1203 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1204 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1205 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1206 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1207 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1208 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1209 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1210 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1211 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1212 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1213 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1214 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1215 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1216 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1217 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1218 Fink, Zbigniew Jędrzejewski-Szmek
1220 — Berlin, 2015-06-19
1224 * The gudev library has been extracted into a separate repository
1225 available at: https://git.gnome.org/browse/libgudev/
1226 It is now managed as part of the Gnome project. Distributions
1227 are recommended to pass --disable-gudev to systemd and use
1228 gudev from the Gnome project instead. gudev is still included
1229 in systemd, for now. It will be removed soon, though. Please
1230 also see the announcement-thread on systemd-devel:
1231 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1233 * systemd now exposes a CPUUsageNSec= property for each
1234 service unit on the bus, that contains the overall consumed
1235 CPU time of a service (the sum of what each process of the
1236 service consumed). This value is only available if
1237 CPUAccounting= is turned on for a service, and is then shown
1238 in the "systemctl status" output.
1240 * Support for configuring alternative mappings of the old SysV
1241 runlevels to systemd targets has been removed. They are now
1242 hardcoded in a way that runlevels 2, 3, 4 all map to
1243 multi-user.target and 5 to graphical.target (which
1244 previously was already the default behaviour).
1246 * The auto-mounter logic gained support for mount point
1247 expiry, using a new TimeoutIdleSec= setting in .automount
1248 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1250 * The EFI System Partition (ESP) as mounted to /boot by
1251 systemd-efi-boot-generator will now be unmounted
1252 automatically after 2 minutes of not being used. This should
1253 minimize the risk of ESP corruptions.
1255 * New /etc/fstab options x-systemd.requires= and
1256 x-systemd.requires-mounts-for= are now supported to express
1257 additional dependencies for mounts. This is useful for
1258 journalling file systems that support external journal
1259 devices or overlay file systems that require underlying file
1260 systems to be mounted.
1262 * systemd does not support direct live-upgrades (via systemctl
1263 daemon-reexec) from versions older than v44 anymore. As no
1264 distribution we are aware of shipped such old versions in a
1265 stable release this should not be problematic.
1267 * When systemd forks off a new per-connection service instance
1268 it will now set the $REMOTE_ADDR environment variable to the
1269 remote IP address, and $REMOTE_PORT environment variable to
1270 the remote IP port. This behaviour is similar to the
1271 corresponding environment variables defined by CGI.
1273 * systemd-networkd gained support for uplink failure
1274 detection. The BindCarrier= option allows binding interface
1275 configuration dynamically to the link sense of other
1276 interfaces. This is useful to achieve behaviour like in
1279 * systemd-networkd gained support for configuring the DHCP
1280 client identifier to use when requesting leases.
1282 * systemd-networkd now has a per-network UseNTP= option to
1283 configure whether NTP server information acquired via DHCP
1284 is passed on to services like systemd-timesyncd.
1286 * systemd-networkd gained support for vti6 tunnels.
1288 * Note that systemd-networkd manages the sysctl variable
1289 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1290 it is configured for since v219. The variable controls IP
1291 forwarding, and is a per-interface alternative to the global
1292 /proc/sys/net/ipv[46]/ip_forward. This setting is
1293 configurable in the IPForward= option, which defaults to
1294 "no". This means if networkd is used for an interface it is
1295 no longer sufficient to set the global sysctl option to turn
1296 on IP forwarding! Instead, the .network file option
1297 IPForward= needs to be turned on! Note that the
1298 implementation of this behaviour was broken in v219 and has
1301 * Many bonding and vxlan options are now configurable in
1304 * systemd-nspawn gained a new --property= setting to set unit
1305 properties for the container scope. This is useful for
1306 setting resource parameters (e.g "CPUShares=500") on
1307 containers started from the command line.
1309 * systemd-nspawn gained a new --private-users= switch to make
1310 use of user namespacing available on recent Linux kernels.
1312 * systemd-nspawn may now be called as part of a shell pipeline
1313 in which case the pipes used for stdin and stdout are passed
1314 directly to the process invoked in the container, without
1315 indirection via a pseudo tty.
1317 * systemd-nspawn gained a new switch to control the UNIX
1318 signal to use when killing the init process of the container
1321 * systemd-nspawn gained a new --overlay= switch for mounting
1322 overlay file systems into the container using the new kernel
1325 * When a container image is imported via systemd-importd and
1326 the host file system is not btrfs, a loopback block device
1327 file is created in /var/lib/machines.raw with a btrfs file
1328 system inside. It is then mounted to /var/lib/machines to
1329 enable btrfs features for container management. The loopback
1330 file and btrfs file system is grown as needed when container
1331 images are imported via systemd-importd.
1333 * systemd-machined/systemd-importd gained support for btrfs
1334 quota, to enforce container disk space limits on disk. This
1335 is exposed in "machinectl set-limit".
1337 * systemd-importd now can import containers from local .tar,
1338 .raw and .qcow2 images, and export them to .tar and .raw. It
1339 can also import dkr v2 images now from the network (on top
1342 * systemd-importd gained support for verifying downloaded
1343 images with gpg2 (previously only gpg1 was supported).
1345 * systemd-machined, systemd-logind, systemd: most bus calls
1346 are now accessible to unprivileged processes via
1347 PolicyKit. Also, systemd-logind will now allow users to kill
1348 their own sessions without further privileges or
1351 * systemd-shutdownd has been removed. This service was
1352 previously responsible for implementing scheduled shutdowns
1353 as exposed in /usr/bin/shutdown's time parameter. This
1354 functionality has now been moved into systemd-logind and is
1355 accessible via a bus interface.
1357 * "systemctl reboot" gained a new switch --firmware-setup that
1358 can be used to reboot into the EFI firmware setup, if that
1359 is available. systemd-logind now exposes an API on the bus
1360 to trigger such reboots, in case graphical desktop UIs want
1361 to cover this functionality.
1363 * "systemctl enable", "systemctl disable" and "systemctl mask"
1364 now support a new "--now" switch. If specified the units
1365 that are enabled will also be started, and the ones
1366 disabled/masked also stopped.
1368 * The Gummiboot EFI boot loader tool has been merged into
1369 systemd, and renamed to "systemd-boot". The bootctl tool has been
1370 updated to support systemd-boot.
1372 * An EFI kernel stub has been added that may be used to create
1373 kernel EFI binaries that contain not only the actual kernel,
1374 but also an initrd, boot splash, command line and OS release
1375 information. This combined binary can then be signed as a
1376 single image, so that the firmware can verify it all in one
1377 step. systemd-boot has special support for EFI binaries created
1378 like this and can extract OS release information from them
1379 and show them in the boot menu. This functionality is useful
1380 to implement cryptographically verified boot schemes.
1382 * Optional support has been added to systemd-fsck to pass
1383 fsck's progress report to an AF_UNIX socket in the file
1386 * udev will no longer create device symlinks for all block
1387 devices by default. A blacklist for excluding special block
1388 devices from this logic has been turned into a whitelist
1389 that requires picking block devices explicitly that require
1392 * A new (currently still internal) API sd-device.h has been
1393 added to libsystemd. This modernized API is supposed to
1394 replace libudev eventually. In fact, already much of libudev
1395 is now just a wrapper around sd-device.h.
1397 * A new hwdb database for storing metadata about pointing
1398 stick devices has been added.
1400 * systemd-tmpfiles gained support for setting file attributes
1401 similar to the "chattr" tool with new 'h' and 'H' lines.
1403 * systemd-journald will no longer unconditionally set the
1404 btrfs NOCOW flag on new journal files. This is instead done
1405 with tmpfiles snippet using the new 'h' line type. This
1406 allows easy disabling of this logic, by masking the
1407 journal-nocow.conf tmpfiles file.
1409 * systemd-journald will now translate audit message types to
1410 human readable identifiers when writing them to the
1411 journal. This should improve readability of audit messages.
1413 * The LUKS logic gained support for the offset= and skip=
1414 options in /etc/crypttab, as previously implemented by
1417 * /usr/lib/os-release gained a new optional field VARIANT= for
1418 distributions that support multiple variants (such as a
1419 desktop edition, a server edition, ...)
1421 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1422 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1423 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1424 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1425 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1426 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1427 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1428 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1429 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1430 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1431 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1432 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1433 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1434 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1435 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1436 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1437 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1438 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1439 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1440 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1441 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1442 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1443 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1444 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1445 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1446 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1447 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1449 — Berlin, 2015-05-22
1453 * Introduce a new API "sd-hwdb.h" for querying the hardware
1454 metadata database. With this minimal interface one can query
1455 and enumerate the udev hwdb, decoupled from the old libudev
1456 library. libudev's interface for this is now only a wrapper
1457 around sd-hwdb. A new tool systemd-hwdb has been added to
1458 interface with and update the database.
1460 * When any of systemd's tools copies files (for example due to
1461 tmpfiles' C lines) a btrfs reflink will attempted first,
1462 before bytewise copying is done.
1464 * systemd-nspawn gained a new --ephemeral switch. When
1465 specified a btrfs snapshot is taken of the container's root
1466 directory, and immediately removed when the container
1467 terminates again. Thus, a container can be started whose
1468 changes never alter the container's root directory, and are
1469 lost on container termination. This switch can also be used
1470 for starting a container off the root file system of the
1471 host without affecting the host OS. This switch is only
1472 available on btrfs file systems.
1474 * systemd-nspawn gained a new --template= switch. It takes the
1475 path to a container tree to use as template for the tree
1476 specified via --directory=, should that directory be
1477 missing. This allows instantiating containers dynamically,
1478 on first run. This switch is only available on btrfs file
1481 * When a .mount unit refers to a mount point on which multiple
1482 mounts are stacked, and the .mount unit is stopped all of
1483 the stacked mount points will now be unmounted until no
1484 mount point remains.
1486 * systemd now has an explicit notion of supported and
1487 unsupported unit types. Jobs enqueued for unsupported unit
1488 types will now fail with an "unsupported" error code. More
1489 specifically .swap, .automount and .device units are not
1490 supported in containers, .busname units are not supported on
1491 non-kdbus systems. .swap and .automount are also not
1492 supported if their respective kernel compile time options
1495 * machinectl gained support for two new "copy-from" and
1496 "copy-to" commands for copying files from a running
1497 container to the host or vice versa.
1499 * machinectl gained support for a new "bind" command to bind
1500 mount host directories into local containers. This is
1501 currently only supported for nspawn containers.
1503 * networkd gained support for configuring bridge forwarding
1504 database entries (fdb) from .network files.
1506 * A new tiny daemon "systemd-importd" has been added that can
1507 download container images in tar, raw, qcow2 or dkr formats,
1508 and make them available locally in /var/lib/machines, so
1509 that they can run as nspawn containers. The daemon can GPG
1510 verify the downloads (not supported for dkr, since it has no
1511 provisions for verifying downloads). It will transparently
1512 decompress bz2, xz, gzip compressed downloads if necessary,
1513 and restore sparse files on disk. The daemon uses privilege
1514 separation to ensure the actual download logic runs with
1515 fewer privileges than the daemon itself. machinectl has
1516 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1517 make the functionality of importd available to the
1518 user. With this in place the Fedora and Ubuntu "Cloud"
1519 images can be downloaded and booted as containers unmodified
1520 (the Fedora images lack the appropriate GPG signature files
1521 currently, so they cannot be verified, but this will change
1522 soon, hopefully). Note that downloading images is currently
1523 only fully supported on btrfs.
1525 * machinectl is now able to list container images found in
1526 /var/lib/machines, along with some metadata about sizes of
1527 disk and similar. If the directory is located on btrfs and
1528 quota is enabled, this includes quota display. A new command
1529 "image-status" has been added that shows additional
1530 information about images.
1532 * machinectl is now able to clone container images
1533 efficiently, if the underlying file system (btrfs) supports
1534 it, with the new "machinectl clone" command. It also
1535 gained commands for renaming and removing images, as well as
1536 marking them read-only or read-write (supported also on
1537 legacy file systems).
1539 * networkd gained support for collecting LLDP network
1540 announcements, from hardware that supports this. This is
1541 shown in networkctl output.
1543 * systemd-run gained support for a new -t (--pty) switch for
1544 invoking a binary on a pty whose input and output is
1545 connected to the invoking terminal. This allows executing
1546 processes as system services while interactively
1547 communicating with them via the terminal. Most interestingly
1548 this is supported across container boundaries. Invoking
1549 "systemd-run -t /bin/bash" is an alternative to running a
1550 full login session, the difference being that the former
1551 will not register a session, nor go through the PAM session
1554 * tmpfiles gained support for a new "v" line type for creating
1555 btrfs subvolumes. If the underlying file system is a legacy
1556 file system, this automatically degrades to creating a
1557 normal directory. Among others /var/lib/machines is now
1558 created like this at boot, should it be missing.
1560 * The directory /var/lib/containers/ has been deprecated and
1561 been replaced by /var/lib/machines. The term "machines" has
1562 been used in the systemd context as generic term for both
1563 VMs and containers, and hence appears more appropriate for
1564 this, as the directory can also contain raw images bootable
1567 * systemd-nspawn when invoked with -M but without --directory=
1568 or --image= is now capable of searching for the container
1569 root directory, subvolume or disk image automatically, in
1570 /var/lib/machines. systemd-nspawn@.service has been updated
1571 to make use of this, thus allowing it to be used for raw
1574 * A new machines.target unit has been introduced that is
1575 supposed to group all containers/VMs invoked as services on
1576 the system. systemd-nspawn@.service has been updated to
1577 integrate with that.
1579 * machinectl gained a new "start" command, for invoking a
1580 container as a service. "machinectl start foo" is mostly
1581 equivalent to "systemctl start systemd-nspawn@foo.service",
1582 but handles escaping in a nicer way.
1584 * systemd-nspawn will now mount most of the cgroupfs tree
1585 read-only into each container, with the exception of the
1586 container's own subtree in the name=systemd hierarchy.
1588 * journald now sets the special FS_NOCOW file flag for its
1589 journal files. This should improve performance on btrfs, by
1590 avoiding heavy fragmentation when journald's write-pattern
1591 is used on COW file systems. It degrades btrfs' data
1592 integrity guarantees for the files to the same levels as for
1593 ext3/ext4 however. This should be OK though as journald does
1594 its own data integrity checks and all its objects are
1595 checksummed on disk. Also, journald should handle btrfs disk
1596 full events a lot more gracefully now, by processing SIGBUS
1597 errors, and not relying on fallocate() anymore.
1599 * When journald detects that journal files it is writing to
1600 have been deleted it will immediately start new journal
1603 * systemd now provides a way to store file descriptors
1604 per-service in PID 1.This is useful for daemons to ensure
1605 that fds they require are not lost during a daemon
1606 restart. The fds are passed to the daemon on the next
1607 invocation in the same way socket activation fds are
1608 passed. This is now used by journald to ensure that the
1609 various sockets connected to all the system's stdout/stderr
1610 are not lost when journald is restarted. File descriptors
1611 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1612 an extension to sd_notify(). Note that a limit is enforced
1613 on the number of fds a service can store in PID 1, and it
1614 defaults to 0, so that no fds may be stored, unless this is
1615 explicitly turned on.
1617 * The default TERM variable to use for units connected to a
1618 terminal, when no other value is explicitly is set is now
1619 vt220 rather than vt102. This should be fairly safe still,
1620 but allows PgUp/PgDn work.
1622 * The /etc/crypttab option header= as known from Debian is now
1625 * "loginctl user-status" and "loginctl session-status" will
1626 now show the last 10 lines of log messages of the
1627 user/session following the status output. Similar,
1628 "machinectl status" will show the last 10 log lines
1629 associated with a virtual machine or container
1630 service. (Note that this is usually not the log messages
1631 done in the VM/container itself, but simply what the
1632 container manager logs. For nspawn this includes all console
1635 * "loginctl session-status" without further argument will now
1636 show the status of the session of the caller. Similar,
1637 "lock-session", "unlock-session", "activate",
1638 "enable-linger", "disable-linger" may now be called without
1639 session/user parameter in which case they apply to the
1640 caller's session/user.
1642 * An X11 session scriptlet is now shipped that uploads
1643 $DISPLAY and $XAUTHORITY into the environment of the systemd
1644 --user daemon if a session begins. This should improve
1645 compatibility with X11 enabled applications run as systemd
1648 * Generators are now subject to masking via /etc and /run, the
1649 same way as unit files.
1651 * networkd .network files gained support for configuring
1652 per-link IPv4/IPv6 packet forwarding as well as IPv4
1653 masquerading. This is by default turned on for veth links to
1654 containers, as registered by systemd-nspawn. This means that
1655 nspawn containers run with --network-veth will now get
1656 automatic routed access to the host's networks without any
1657 further configuration or setup, as long as networkd runs on
1660 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1661 or UDP posts of a container on the host. With this in place
1662 it is possible to run containers with private veth links
1663 (--network-veth), and have their functionality exposed on
1664 the host as if their services were running directly on the
1667 * systemd-nspawn's --network-veth switch now gained a short
1668 version "-n", since with the changes above it is now truly
1669 useful out-of-the-box. The systemd-nspawn@.service has been
1670 updated to make use of it too by default.
1672 * systemd-nspawn will now maintain a per-image R/W lock, to
1673 ensure that the same image is not started more than once
1674 writable. (It's OK to run an image multiple times
1675 simultaneously in read-only mode.)
1677 * systemd-nspawn's --image= option is now capable of
1678 dissecting and booting MBR and GPT disk images that contain
1679 only a single active Linux partition. Previously it
1680 supported only GPT disk images with proper GPT type
1681 IDs. This allows running cloud images from major
1682 distributions directly with systemd-nspawn, without
1685 * In addition to collecting mouse dpi data in the udev
1686 hardware database, there's now support for collecting angle
1687 information for mouse scroll wheels. The database is
1688 supposed to guarantee similar scrolling behavior on mice
1689 that it knows about. There's also support for collecting
1690 information about Touchpad types.
1692 * udev's input_id built-in will now also collect touch screen
1693 dimension data and attach it to probed devices.
1695 * /etc/os-release gained support for a Distribution Privacy
1698 * networkd gained support for creating "ipvlan", "gretap",
1699 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1701 * systemd-tmpfiles gained support for "a" lines for setting
1704 * systemd-nspawn will now mount /tmp in the container to
1705 tmpfs, automatically.
1707 * systemd now exposes the memory.usage_in_bytes cgroup
1708 attribute and shows it for each service in the "systemctl
1709 status" output, if available.
1711 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1712 immediate reboot is triggered. This useful if shutdown is
1713 hung and is unable to complete, to expedite the
1714 operation. Note that this kind of reboot will still unmount
1715 all file systems, and hence should not result in fsck being
1718 * A .device unit for an optical block device will now be
1719 considered active only when a medium is in the drive. Also,
1720 mount units are now bound to their backing devices thus
1721 triggering automatic unmounting when devices become
1722 unavailable. With this in place systemd will now
1723 automatically unmount left-over mounts when a CD-ROM is
1724 ejected or an USB stick is yanked from the system.
1726 * networkd-wait-online now has support for waiting for
1727 specific interfaces only (with globbing), and for giving up
1728 after a configurable timeout.
1730 * networkd now exits when idle. It will be automatically
1731 restarted as soon as interfaces show up, are removed or
1732 change state. networkd will stay around as long as there is
1733 at least one DHCP state machine or similar around, that keep
1736 * networkd may now configure IPv6 link-local addressing in
1737 addition to IPv4 link-local addressing.
1739 * The IPv6 "token" for use in SLAAC may now be configured for
1740 each .network interface in networkd.
1742 * Routes configured with networkd may now be assigned a scope
1745 * networkd's [Match] sections now support globbing and lists
1746 of multiple space-separated matches per item.
1748 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1749 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1750 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1751 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1752 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1753 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1754 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1755 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1756 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1757 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1758 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1759 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1760 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1761 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1762 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1763 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1764 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1765 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1766 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1767 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1768 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1769 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1770 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1771 Hoffmann, Zbigniew Jędrzejewski-Szmek
1773 — Berlin, 2015-02-16
1777 * When querying unit file enablement status (for example via
1778 "systemctl is-enabled"), a new state "indirect" is now known
1779 which indicates that a unit might not be enabled itself, but
1780 another unit listed in its Also= setting might be.
1782 * Similar to the various existing ConditionXYZ= settings for
1783 units, there are now matching AssertXYZ= settings. While
1784 failing conditions cause a unit to be skipped, but its job
1785 to succeed, failing assertions declared like this will cause
1786 a unit start operation and its job to fail.
1788 * hostnamed now knows a new chassis type "embedded".
1790 * systemctl gained a new "edit" command. When used on a unit
1791 file, this allows extending unit files with .d/ drop-in
1792 configuration snippets or editing the full file (after
1793 copying it from /usr/lib to /etc). This will invoke the
1794 user's editor (as configured with $EDITOR), and reload the
1795 modified configuration after editing.
1797 * "systemctl status" now shows the suggested enablement state
1798 for a unit, as declared in the (usually vendor-supplied)
1799 system preset files.
1801 * nss-myhostname will now resolve the single-label host name
1802 "gateway" to the locally configured default IP routing
1803 gateways, ordered by their metrics. This assigns a stable
1804 name to the used gateways, regardless which ones are
1805 currently configured. Note that the name will only be
1806 resolved after all other name sources (if nss-myhostname is
1807 configured properly) and should hence not negatively impact
1808 systems that use the single-label host name "gateway" in
1811 * systemd-inhibit now allows filtering by mode when listing
1814 * Scope and service units gained a new "Delegate" boolean
1815 property, which, when set, allows processes running inside the
1816 unit to further partition resources. This is primarily
1817 useful for systemd user instances as well as container
1820 * journald will now pick up audit messages directly from
1821 the kernel, and log them like any other log message. The
1822 audit fields are split up and fully indexed. This means that
1823 journalctl in many ways is now a (nicer!) alternative to
1824 ausearch, the traditional audit client. Note that this
1825 implements only a minimal audit client. If you want the
1826 special audit modes like reboot-on-log-overflow, please use
1827 the traditional auditd instead, which can be used in
1828 parallel to journald.
1830 * The ConditionSecurity= unit file option now understands the
1831 special string "audit" to check whether auditing is
1834 * journalctl gained two new commands --vacuum-size= and
1835 --vacuum-time= to delete old journal files until the
1836 remaining ones take up no more than the specified size on disk,
1837 or are not older than the specified time.
1839 * A new, native PPPoE library has been added to sd-network,
1840 systemd's library of light-weight networking protocols. This
1841 library will be used in a future version of networkd to
1842 enable PPPoE communication without an external pppd daemon.
1844 * The busctl tool now understands a new "capture" verb that
1845 works similar to "monitor", but writes a packet capture
1846 trace to STDOUT that can be redirected to a file which is
1847 compatible with libcap's capture file format. This can then
1848 be loaded in Wireshark and similar tools to inspect bus
1851 * The busctl tool now understands a new "tree" verb that shows
1852 the object trees of a specific service on the bus, or of all
1855 * The busctl tool now understands a new "introspect" verb that
1856 shows all interfaces and members of objects on the bus,
1857 including their signature and values. This is particularly
1858 useful to get more information about bus objects shown by
1859 the new "busctl tree" command.
1861 * The busctl tool now understands new verbs "call",
1862 "set-property" and "get-property" for invoking bus method
1863 calls, setting and getting bus object properties in a
1866 * busctl gained a new --augment-creds= argument that controls
1867 whether the tool shall augment credential information it
1868 gets from the bus with data from /proc, in a possibly
1871 * nspawn's --link-journal= switch gained two new values
1872 "try-guest" and "try-host" that work like "guest" and
1873 "host", but do not fail if the host has no persistent
1874 journalling enabled. -j is now equivalent to
1875 --link-journal=try-guest.
1877 * macvlan network devices created by nspawn will now have
1878 stable MAC addresses.
1880 * A new SmackProcessLabel= unit setting has been added, which
1881 controls the SMACK security label processes forked off by
1882 the respective unit shall use.
1884 * If compiled with --enable-xkbcommon, systemd-localed will
1885 verify x11 keymap settings by compiling the given keymap. It
1886 will spew out warnings if the compilation fails. This
1887 requires libxkbcommon to be installed.
1889 * When a coredump is collected, a larger number of metadata
1890 fields is now collected and included in the journal records
1891 created for it. More specifically, control group membership,
1892 environment variables, memory maps, working directory,
1893 chroot directory, /proc/$PID/status, and a list of open file
1894 descriptors is now stored in the log entry.
1896 * The udev hwdb now contains DPI information for mice. For
1899 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1901 * All systemd programs that read standalone configuration
1902 files in /etc now also support a corresponding series of
1903 .conf.d configuration directories in /etc/, /run/,
1904 /usr/local/lib/, /usr/lib/, and (if configured with
1905 --enable-split-usr) /lib/. In particular, the following
1906 configuration files now have corresponding configuration
1907 directories: system.conf user.conf, logind.conf,
1908 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1909 resolved.conf, timesyncd.conf, journal-remote.conf, and
1910 journal-upload.conf. Note that distributions should use the
1911 configuration directories in /usr/lib/; the directories in
1912 /etc/ are reserved for the system administrator.
1914 * systemd-rfkill will no longer take the rfkill device name
1915 into account when storing rfkill state on disk, as the name
1916 might be dynamically assigned and not stable. Instead, the
1917 ID_PATH udev variable combined with the rfkill type (wlan,
1918 bluetooth, ...) is used.
1920 * A new service systemd-machine-id-commit.service has been
1921 added. When used on systems where /etc is read-only during
1922 boot, and /etc/machine-id is not initialized (but an empty
1923 file), this service will copy the temporary machine ID
1924 created as replacement into /etc after the system is fully
1925 booted up. This is useful for systems that are freshly
1926 installed with a non-initialized machine ID, but should get
1927 a fixed machine ID for subsequent boots.
1929 * networkd's .netdev files now provide a large set of
1930 configuration parameters for VXLAN devices. Similarly, the
1931 bridge port cost parameter is now configurable in .network
1932 files. There's also new support for configuring IP source
1933 routing. networkd .link files gained support for a new
1934 OriginalName= match that is useful to match against the
1935 original interface name the kernel assigned. .network files
1936 may include MTU= and MACAddress= fields for altering the MTU
1937 and MAC address while being connected to a specific network
1940 * The LUKS logic gained supported for configuring
1941 UUID-specific key files. There's also new support for naming
1942 LUKS device from the kernel command line, using the new
1943 luks.name= argument.
1945 * Timer units may now be transiently created via the bus API
1946 (this was previously already available for scope and service
1947 units). In addition it is now possible to create multiple
1948 transient units at the same time with a single bus call. The
1949 "systemd-run" tool has been updated to make use of this for
1950 running commands on a specified time, in at(1)-style.
1952 * tmpfiles gained support for "t" lines, for assigning
1953 extended attributes to files. Among other uses this may be
1954 used to assign SMACK labels to files.
1956 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1957 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1958 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1959 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1960 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1961 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1962 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1963 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1964 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1965 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1966 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1967 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1968 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1969 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1970 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1971 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1972 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1973 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1975 — Berlin, 2014-12-10
1979 * journalctl gained the new options -t/--identifier= to match
1980 on the syslog identifier (aka "tag"), as well as --utc to
1981 show log timestamps in the UTC timezone. journalctl now also
1982 accepts -n/--lines=all to disable line capping in a pager.
1984 * journalctl gained a new switch, --flush, that synchronously
1985 flushes logs from /run/log/journal to /var/log/journal if
1986 persistent storage is enabled. systemd-journal-flush.service
1987 now waits until the operation is complete.
1989 * Services can notify the manager before they start a reload
1990 (by sending RELOADING=1) or shutdown (by sending
1991 STOPPING=1). This allows the manager to track and show the
1992 internal state of daemons and closes a race condition when
1993 the process is still running but has closed its D-Bus
1996 * Services with Type=oneshot do not have to have any ExecStart
1999 * User units are now loaded also from
2000 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2001 /run/systemd/user directory that was already previously
2002 supported, but is under the control of the user.
2004 * Job timeouts (i.e. time-outs on the time a job that is
2005 queued stays in the run queue) can now optionally result in
2006 immediate reboot or power-off actions (JobTimeoutAction= and
2007 JobTimeoutRebootArgument=). This is useful on ".target"
2008 units, to limit the maximum time a target remains
2009 undispatched in the run queue, and to trigger an emergency
2010 operation in such a case. This is now used by default to
2011 turn off the system if boot-up (as defined by everything in
2012 basic.target) hangs and does not complete for at least
2013 15min. Also, if power-off or reboot hang for at least 30min
2014 an immediate power-off/reboot operation is triggered. This
2015 functionality is particularly useful to increase reliability
2016 on embedded devices, but also on laptops which might
2017 accidentally get powered on when carried in a backpack and
2018 whose boot stays stuck in a hard disk encryption passphrase
2021 * systemd-logind can be configured to also handle lid switch
2022 events even when the machine is docked or multiple displays
2023 are attached (HandleLidSwitchDocked= option).
2025 * A helper binary and a service have been added which can be
2026 used to resume from hibernation in the initramfs. A
2027 generator will parse the resume= option on the kernel
2028 command line to trigger resume.
2030 * A user console daemon systemd-consoled has been
2031 added. Currently, it is a preview, and will so far open a
2032 single terminal on each session of the user marked as
2033 Desktop=systemd-console.
2035 * Route metrics can be specified for DHCP routes added by
2038 * The SELinux context of socket-activated services can be set
2039 from the information provided by the networking stack
2040 (SELinuxContextFromNet= option).
2042 * Userspace firmware loading support has been removed and
2043 the minimum supported kernel version is thus bumped to 3.7.
2045 * Timeout for udev workers has been increased from 1 to 3
2046 minutes, but a warning will be printed after 1 minute to
2047 help diagnose kernel modules that take a long time to load.
2049 * Udev rules can now remove tags on devices with TAG-="foobar".
2051 * systemd's readahead implementation has been removed. In many
2052 circumstances it didn't give expected benefits even for
2053 rotational disk drives and was becoming less relevant in the
2054 age of SSDs. As none of the developers has been using
2055 rotating media anymore, and nobody stepped up to actively
2056 maintain this component of systemd it has now been removed.
2058 * Swap units can use Options= to specify discard options.
2059 Discard options specified for swaps in /etc/fstab are now
2062 * Docker containers are now detected as a separate type of
2065 * The Password Agent protocol gained support for queries where
2066 the user input is shown, useful e.g. for user names.
2067 systemd-ask-password gained a new --echo option to turn that
2070 * The default sysctl.d/ snippets will now set:
2072 net.core.default_qdisc = fq_codel
2074 This selects Fair Queuing Controlled Delay as the default
2075 queuing discipline for network interfaces. fq_codel helps
2076 fight the network bufferbloat problem. It is believed to be
2077 a good default with no tuning required for most workloads.
2078 Downstream distributions may override this choice. On 10Gbit
2079 servers that do not do forwarding, "fq" may perform better.
2080 Systems without a good clocksource should use "pfifo_fast".
2082 * If kdbus is enabled during build a new option BusPolicy= is
2083 available for service units, that allows locking all service
2084 processes into a stricter bus policy, in order to limit
2085 access to various bus services, or even hide most of them
2086 from the service's view entirely.
2088 * networkctl will now show the .network and .link file
2089 networkd has applied to a specific interface.
2091 * sd-login gained a new API call sd_session_get_desktop() to
2092 query which desktop environment has been selected for a
2095 * UNIX utmp support is now compile-time optional to support
2096 legacy-free systems.
2098 * systemctl gained two new commands "add-wants" and
2099 "add-requires" for pulling in units from specific targets
2102 * If the word "rescue" is specified on the kernel command line
2103 the system will now boot into rescue mode (aka
2104 rescue.target), which was previously available only by
2105 specifying "1" or "systemd.unit=rescue.target" on the kernel
2106 command line. This new kernel command line option nicely
2107 mirrors the already existing "emergency" kernel command line
2110 * New kernel command line options mount.usr=, mount.usrflags=,
2111 mount.usrfstype= have been added that match root=, rootflags=,
2112 rootfstype= but allow mounting a specific file system to
2115 * The $NOTIFY_SOCKET is now also passed to control processes of
2116 services, not only the main process.
2118 * This version reenables support for fsck's -l switch. This
2119 means at least version v2.25 of util-linux is required for
2120 operation, otherwise dead-locks on device nodes may
2121 occur. Again: you need to update util-linux to at least
2122 v2.25 when updating systemd to v217.
2124 * The "multi-seat-x" tool has been removed from systemd, as
2125 its functionality has been integrated into X servers 1.16,
2126 and the tool is hence redundant. It is recommended to update
2127 display managers invoking this tool to simply invoke X
2128 directly from now on, again.
2130 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2131 message flag has been added for all of systemd's PolicyKit
2132 authenticated method calls has been added. In particular
2133 this now allows optional interactive authorization via
2134 PolicyKit for many of PID1's privileged operations such as
2135 unit file enabling and disabling.
2137 * "udevadm hwdb --update" learnt a new switch "--usr" for
2138 placing the rebuilt hardware database in /usr instead of
2139 /etc. When used only hardware database entries stored in
2140 /usr will be used, and any user database entries in /etc are
2141 ignored. This functionality is useful for vendors to ship a
2142 pre-built database on systems where local configuration is
2143 unnecessary or unlikely.
2145 * Calendar time specifications in .timer units now also
2146 understand the strings "semi-annually", "quarterly" and
2147 "minutely" as shortcuts (in addition to the preexisting
2148 "anually", "hourly", ...).
2150 * systemd-tmpfiles will now correctly create files in /dev
2151 at boot which are marked for creation only at boot. It is
2152 recommended to always create static device nodes with 'c!'
2153 and 'b!', so that they are created only at boot and not
2154 overwritten at runtime.
2156 * When the watchdog logic is used for a service (WatchdogSec=)
2157 and the watchdog timeout is hit the service will now be
2158 terminated with SIGABRT (instead of just SIGTERM), in order
2159 to make sure a proper coredump and backtrace is
2160 generated. This ensures that hanging services will result in
2161 similar coredump/backtrace behaviour as services that hit a
2164 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2165 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2166 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2167 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2168 Herrmann, David Sommerseth, David Strauss, Emil Renner
2169 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2170 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2171 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2172 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2173 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2174 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2175 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2176 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2177 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2178 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2179 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2180 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2181 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2182 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2183 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2184 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2187 — Berlin, 2014-10-28
2191 * timedated no longer reads NTP implementation unit names from
2192 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2193 implementations should add a
2195 Conflicts=systemd-timesyncd.service
2197 to their unit files to take over and replace systemd's NTP
2198 default functionality.
2200 * systemd-sysusers gained a new line type "r" for configuring
2201 which UID/GID ranges to allocate system users/groups
2202 from. Lines of type "u" may now add an additional column
2203 that specifies the home directory for the system user to be
2204 created. Also, systemd-sysusers may now optionally read user
2205 information from STDIN instead of a file. This is useful for
2206 invoking it from RPM preinst scriptlets that need to create
2207 users before the first RPM file is installed since these
2208 files might need to be owned by them. A new
2209 %sysusers_create_inline RPM macro has been introduced to do
2210 just that. systemd-sysusers now updates the shadow files as
2211 well as the user/group databases, which should enhance
2212 compatibility with certain tools like grpck.
2214 * A number of bus APIs of PID 1 now optionally consult
2215 PolicyKit to permit access for otherwise unprivileged
2216 clients under certain conditions. Note that this currently
2217 doesn't support interactive authentication yet, but this is
2218 expected to be added eventually, too.
2220 * /etc/machine-info now has new fields for configuring the
2221 deployment environment of the machine, as well as the
2222 location of the machine. hostnamectl has been updated with
2223 new command to update these fields.
2225 * systemd-timesyncd has been updated to automatically acquire
2226 NTP server information from systemd-networkd, which might
2227 have been discovered via DHCP.
2229 * systemd-resolved now includes a caching DNS stub resolver
2230 and a complete LLMNR name resolution implementation. A new
2231 NSS module "nss-resolve" has been added which can be used
2232 instead of glibc's own "nss-dns" to resolve hostnames via
2233 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2234 be resolved via systemd-resolved D-Bus APIs. In contrast to
2235 the glibc internal resolver systemd-resolved is aware of
2236 multi-homed system, and keeps DNS server and caches separate
2237 and per-interface. Queries are sent simultaneously on all
2238 interfaces that have DNS servers configured, in order to
2239 properly handle VPNs and local LANs which might resolve
2240 separate sets of domain names. systemd-resolved may acquire
2241 DNS server information from systemd-networkd automatically,
2242 which in turn might have discovered them via DHCP. A tool
2243 "systemd-resolve-host" has been added that may be used to
2244 query the DNS logic in resolved. systemd-resolved implements
2245 IDNA and automatically uses IDNA or UTF-8 encoding depending
2246 on whether classic DNS or LLMNR is used as transport. In the
2247 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2248 implementation to systemd-resolved.
2250 * A new NSS module nss-mymachines has been added, that
2251 automatically resolves the names of all local registered
2252 containers to their respective IP addresses.
2254 * A new client tool "networkctl" for systemd-networkd has been
2255 added. It currently is entirely passive and will query
2256 networking configuration from udev, rtnetlink and networkd,
2257 and present it to the user in a very friendly
2258 way. Eventually, we hope to extend it to become a full
2259 control utility for networkd.
2261 * .socket units gained a new DeferAcceptSec= setting that
2262 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2263 TCP. Similarly, support for controlling TCP keep-alive
2264 settings has been added (KeepAliveTimeSec=,
2265 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2266 turning off Nagle's algorithm on TCP has been added
2269 * logind learned a new session type "web", for use in projects
2270 like Cockpit which register web clients as PAM sessions.
2272 * timer units with at least one OnCalendar= setting will now
2273 be started only after timer-sync.target has been
2274 reached. This way they will not elapse before the system
2275 clock has been corrected by a local NTP client or
2276 similar. This is particular useful on RTC-less embedded
2277 machines, that come up with an invalid system clock.
2279 * systemd-nspawn's --network-veth= switch should now result in
2280 stable MAC addresses for both the outer and the inner side
2283 * systemd-nspawn gained a new --volatile= switch for running
2284 container instances with /etc or /var unpopulated.
2286 * The kdbus client code has been updated to use the new Linux
2287 3.17 memfd subsystem instead of the old kdbus-specific one.
2289 * systemd-networkd's DHCP client and server now support
2290 FORCERENEW. There are also new configuration options to
2291 configure the vendor client identifier and broadcast mode
2294 * systemd will no longer inform the kernel about the current
2295 timezone, as this is necessarily incorrect and racy as the
2296 kernel has no understanding of DST and similar
2297 concepts. This hence means FAT timestamps will be always
2298 considered UTC, similar to what Android is already
2299 doing. Also, when the RTC is configured to the local time
2300 (rather than UTC) systemd will never synchronize back to it,
2301 as this might confuse Windows at a later boot.
2303 * systemd-analyze gained a new command "verify" for offline
2304 validation of unit files.
2306 * systemd-networkd gained support for a couple of additional
2307 settings for bonding networking setups. Also, the metric for
2308 statically configured routes may now be configured. For
2309 network interfaces where this is appropriate the peer IP
2310 address may now be configured.
2312 * systemd-networkd's DHCP client will no longer request
2313 broadcasting by default, as this tripped up some networks.
2314 For hardware where broadcast is required the feature should
2315 be switched back on using RequestBroadcast=yes.
2317 * systemd-networkd will now set up IPv4LL addresses (when
2318 enabled) even if DHCP is configured successfully.
2320 * udev will now default to respect network device names given
2321 by the kernel when the kernel indicates that these are
2322 predictable. This behavior can be tweaked by changing
2323 NamePolicy= in the relevant .link file.
2325 * A new library systemd-terminal has been added that
2326 implements full TTY stream parsing and rendering. This
2327 library is supposed to be used later on for implementing a
2328 full userspace VT subsystem, replacing the current kernel
2331 * A new tool systemd-journal-upload has been added to push
2332 journal data to a remote system running
2333 systemd-journal-remote.
2335 * journald will no longer forward all local data to another
2336 running syslog daemon. This change has been made because
2337 rsyslog (which appears to be the most commonly used syslog
2338 implementation these days) no longer makes use of this, and
2339 instead pulls the data out of the journal on its own. Since
2340 forwarding the messages to a non-existent syslog server is
2341 more expensive than we assumed we have now turned this
2342 off. If you run a syslog server that is not a recent rsyslog
2343 version, you have to turn this option on again
2344 (ForwardToSyslog= in journald.conf).
2346 * journald now optionally supports the LZ4 compressor for
2347 larger journal fields. This compressor should perform much
2348 better than XZ which was the previous default.
2350 * machinectl now shows the IP addresses of local containers,
2351 if it knows them, plus the interface name of the container.
2353 * A new tool "systemd-escape" has been added that makes it
2354 easy to escape strings to build unit names and similar.
2356 * sd_notify() messages may now include a new ERRNO= field
2357 which is parsed and collected by systemd and shown among the
2358 "systemctl status" output for a service.
2360 * A new component "systemd-firstboot" has been added that
2361 queries the most basic systemd information (timezone,
2362 hostname, root password) interactively on first
2363 boot. Alternatively it may also be used to provision these
2364 things offline on OS images installed into directories.
2366 * The default sysctl.d/ snippets will now set
2368 net.ipv4.conf.default.promote_secondaries=1
2370 This has the benefit of no flushing secondary IP addresses
2371 when primary addresses are removed.
2373 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2374 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2375 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2376 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2377 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2378 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2379 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2380 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2381 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2382 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2383 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2384 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2385 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2386 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2387 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2389 — Berlin, 2014-08-19
2393 * A new tool systemd-sysusers has been added. This tool
2394 creates system users and groups in /etc/passwd and
2395 /etc/group, based on static declarative system user/group
2396 definitions in /usr/lib/sysusers.d/. This is useful to
2397 enable factory resets and volatile systems that boot up with
2398 an empty /etc directory, and thus need system users and
2399 groups created during early boot. systemd now also ships
2400 with two default sysusers.d/ files for the most basic
2401 users and groups systemd and the core operating system
2404 * A new tmpfiles snippet has been added that rebuilds the
2405 essential files in /etc on boot, should they be missing.
2407 * A directive for ensuring automatic clean-up of
2408 /var/cache/man/ has been removed from the default
2409 configuration. This line should now be shipped by the man
2410 implementation. The necessary change has been made to the
2411 man-db implementation. Note that you need to update your man
2412 implementation to one that ships this line, otherwise no
2413 automatic clean-up of /var/cache/man will take place.
2415 * A new condition ConditionNeedsUpdate= has been added that
2416 may conditionalize services to only run when /etc or /var
2417 are "older" than the vendor operating system resources in
2418 /usr. This is useful for reconstructing or updating /etc
2419 after an offline update of /usr or a factory reset, on the
2420 next reboot. Services that want to run once after such an
2421 update or reset should use this condition and order
2422 themselves before the new systemd-update-done.service, which
2423 will mark the two directories as fully updated. A number of
2424 service files have been added making use of this, to rebuild
2425 the udev hardware database, the journald message catalog and
2426 dynamic loader cache (ldconfig). The systemd-sysusers tool
2427 described above also makes use of this now. With this in
2428 place it is now possible to start up a minimal operating
2429 system with /etc empty cleanly. For more information on the
2430 concepts involved see this recent blog story:
2432 http://0pointer.de/blog/projects/stateless.html
2434 * A new system group "input" has been introduced, and all
2435 input device nodes get this group assigned. This is useful
2436 for system-level software to get access to input devices. It
2437 complements what is already done for "audio" and "video".
2439 * systemd-networkd learnt minimal DHCPv4 server support in
2440 addition to the existing DHCPv4 client support. It also
2441 learnt DHCPv6 client and IPv6 Router Solicitation client
2442 support. The DHCPv4 client gained support for static routes
2443 passed in from the server. Note that the [DHCPv4] section
2444 known in older systemd-networkd versions has been renamed to
2445 [DHCP] and is now also used by the DHCPv6 client. Existing
2446 .network files using settings of this section should be
2447 updated, though compatibility is maintained. Optionally, the
2448 client hostname may now be sent to the DHCP server.
2450 * networkd gained support for vxlan virtual networks as well
2451 as tun/tap and dummy devices.
2453 * networkd gained support for automatic allocation of address
2454 ranges for interfaces from a system-wide pool of
2455 addresses. This is useful for dynamically managing a large
2456 number of interfaces with a single network configuration
2457 file. In particular this is useful to easily assign
2458 appropriate IP addresses to the veth links of a large number
2459 of nspawn instances.
2461 * RPM macros for processing sysusers, sysctl and binfmt
2462 drop-in snippets at package installation time have been
2465 * The /etc/os-release file should now be placed in
2466 /usr/lib/os-release. The old location is automatically
2467 created as symlink. /usr/lib is the more appropriate
2468 location of this file, since it shall actually describe the
2469 vendor operating system shipped in /usr, and not the
2470 configuration stored in /etc.
2472 * .mount units gained a new boolean SloppyOptions= setting
2473 that maps to mount(8)'s -s option which enables permissive
2474 parsing of unknown mount options.
2476 * tmpfiles learnt a new "L+" directive which creates a symlink
2477 but (unlike "L") deletes a pre-existing file first, should
2478 it already exist and not already be the correct
2479 symlink. Similarly, "b+", "c+" and "p+" directives have been
2480 added as well, which create block and character devices, as
2481 well as fifos in the filesystem, possibly removing any
2482 pre-existing files of different types.
2484 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2485 'argument' field (which so far specified the source to
2486 symlink/copy the files from) is now optional. If omitted the
2487 same file os copied from /usr/share/factory/ suffixed by the
2488 full destination path. This is useful for populating /etc
2489 with essential files, by copying them from vendor defaults
2490 shipped in /usr/share/factory/etc.
2492 * A new command "systemctl preset-all" has been added that
2493 applies the service preset settings to all installed unit
2494 files. A new switch --preset-mode= has been added that
2495 controls whether only enable or only disable operations
2498 * A new command "systemctl is-system-running" has been added
2499 that allows checking the overall state of the system, for
2500 example whether it is fully up and running.
2502 * When the system boots up with an empty /etc, the equivalent
2503 to "systemctl preset-all" is executed during early boot, to
2504 make sure all default services are enabled after a factory
2507 * systemd now contains a minimal preset file that enables the
2508 most basic services systemd ships by default.
2510 * Unit files' [Install] section gained a new DefaultInstance=
2511 field for defining the default instance to create if a
2512 template unit is enabled with no instance specified.
2514 * A new passive target cryptsetup-pre.target has been added
2515 that may be used by services that need to make they run and
2516 finish before the first LUKS cryptographic device is set up.
2518 * The /dev/loop-control and /dev/btrfs-control device nodes
2519 are now owned by the "disk" group by default, opening up
2520 access to this group.
2522 * systemd-coredump will now automatically generate a
2523 stack trace of all core dumps taking place on the system,
2524 based on elfutils' libdw library. This stack trace is logged
2527 * systemd-coredump may now optionally store coredumps directly
2528 on disk (in /var/lib/systemd/coredump, possibly compressed),
2529 instead of storing them unconditionally in the journal. This
2530 mode is the new default. A new configuration file
2531 /etc/systemd/coredump.conf has been added to configure this
2532 and other parameters of systemd-coredump.
2534 * coredumpctl gained a new "info" verb to show details about a
2535 specific coredump. A new switch "-1" has also been added
2536 that makes sure to only show information about the most
2537 recent entry instead of all entries. Also, as the tool is
2538 generally useful now the "systemd-" prefix of the binary
2539 name has been removed. Distributions that want to maintain
2540 compatibility with the old name should add a symlink from
2541 the old name to the new name.
2543 * journald's SplitMode= now defaults to "uid". This makes sure
2544 that unprivileged users can access their own coredumps with
2545 coredumpctl without restrictions.
2547 * New kernel command line options "systemd.wants=" (for
2548 pulling an additional unit during boot), "systemd.mask="
2549 (for masking a specific unit for the boot), and
2550 "systemd.debug-shell" (for enabling the debug shell on tty9)
2551 have been added. This is implemented in the new generator
2552 "systemd-debug-generator".
2554 * systemd-nspawn will now by default filter a couple of
2555 syscalls for containers, among them those required for
2556 kernel module loading, direct x86 IO port access, swap
2557 management, and kexec. Most importantly though
2558 open_by_handle_at() is now prohibited for containers,
2559 closing a hole similar to a recently discussed vulnerability
2560 in docker regarding access to files on file hierarchies the
2561 container should normally not have access to. Note that, for
2562 nspawn, we generally make no security claims anyway (and
2563 this is explicitly documented in the man page), so this is
2564 just a fix for one of the most obvious problems.
2566 * A new man page file-hierarchy(7) has been added that
2567 contains a minimized, modernized version of the file system
2568 layout systemd expects, similar in style to the FHS
2569 specification or hier(5). A new tool systemd-path(1) has
2570 been added to query many of these paths for the local
2573 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2574 longer done. Since the directory now has a per-user size
2575 limit, and is cleaned on logout this appears unnecessary,
2576 in particular since this now brings the lifecycle of this
2577 directory closer in line with how IPC objects are handled.
2579 * systemd.pc now exports a number of additional directories,
2580 including $libdir (which is useful to identify the library
2581 path for the primary architecture of the system), and a
2582 couple of drop-in directories.
2584 * udev's predictable network interface names now use the dev_port
2585 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2586 distinguish between ports of the same PCI function. dev_id should
2587 only be used for ports using the same HW address, hence the need
2590 * machined has been updated to export the OS version of a
2591 container (read from /etc/os-release and
2592 /usr/lib/os-release) on the bus. This is now shown in
2593 "machinectl status" for a machine.
2595 * A new service setting RestartForceExitStatus= has been
2596 added. If configured to a set of exit signals or process
2597 return values, the service will be restarted when the main
2598 daemon process exits with any of them, regardless of the
2601 * systemctl's -H switch for connecting to remote systemd
2602 machines has been extended so that it may be used to
2603 directly connect to a specific container on the
2604 host. "systemctl -H root@foobar:waldi" will now connect as
2605 user "root" to host "foobar", and then proceed directly to
2606 the container named "waldi". Note that currently you have to
2607 authenticate as user "root" for this to work, as entering
2608 containers is a privileged operation.
2610 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2611 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2612 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2613 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2614 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2615 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2616 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2617 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2618 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2619 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2620 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2621 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2623 — Berlin, 2014-07-03
2627 * As an experimental feature, udev now tries to lock the
2628 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2629 executes events for the disk or any of its partitions.
2630 Applications like partitioning programs can lock the
2631 disk device node (flock(LOCK_EX)) and claim temporary
2632 device ownership that way; udev will entirely skip all event
2633 handling for this disk and its partitions. If the disk
2634 was opened for writing, the close will trigger a partition
2635 table rescan in udev's "watch" facility, and if needed
2636 synthesize "change" events for the disk and all its partitions.
2637 This is now unconditionally enabled, and if it turns out to
2638 cause major problems, we might turn it on only for specific
2639 devices, or might need to disable it entirely. Device Mapper
2640 devices are excluded from this logic.
2642 * We temporarily dropped the "-l" switch for fsck invocations,
2643 since they collide with the flock() logic above. util-linux
2644 upstream has been changed already to avoid this conflict,
2645 and we will readd "-l" as soon as util-linux with this
2646 change has been released.
2648 * The dependency on libattr has been removed. Since a long
2649 time, the extended attribute calls have moved to glibc, and
2650 libattr is thus unnecessary.
2652 * Virtualization detection works without priviliges now. This
2653 means the systemd-detect-virt binary no longer requires
2654 CAP_SYS_PTRACE file capabilities, and our daemons can run
2655 with fewer privileges.
2657 * systemd-networkd now runs under its own "systemd-network"
2658 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2659 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2660 loses the ability to write to files owned by root this way.
2662 * Similarly, systemd-resolved now runs under its own
2663 "systemd-resolve" user with no capabilities remaining.
2665 * Similarly, systemd-bus-proxyd now runs under its own
2666 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2668 * systemd-networkd gained support for setting up "veth"
2669 virtual Ethernet devices for container connectivity, as well
2670 as GRE and VTI tunnels.
2672 * systemd-networkd will no longer automatically attempt to
2673 manually load kernel modules necessary for certain tunnel
2674 transports. Instead, it is assumed the kernel loads them
2675 automatically when required. This only works correctly on
2676 very new kernels. On older kernels, please consider adding
2677 the kernel modules to /etc/modules-load.d/ as a work-around.
2679 * The resolv.conf file systemd-resolved generates has been
2680 moved to /run/systemd/resolve/. If you have a symlink from
2681 /etc/resolv.conf, it might be necessary to correct it.
2683 * Two new service settings, ProtectHome= and ProtectSystem=,
2684 have been added. When enabled, they will make the user data
2685 (such as /home) inaccessible or read-only and the system
2686 (such as /usr) read-only, for specific services. This allows
2687 very light-weight per-service sandboxing to avoid
2688 modifications of user data or system files from
2689 services. These two new switches have been enabled for all
2690 of systemd's long-running services, where appropriate.
2692 * Socket units gained new SocketUser= and SocketGroup=
2693 settings to set the owner user and group of AF_UNIX sockets
2694 and FIFOs in the file system.
2696 * Socket units gained a new RemoveOnStop= setting. If enabled,
2697 all FIFOS and sockets in the file system will be removed
2698 when the specific socket unit is stopped.
2700 * Socket units gained a new Symlinks= setting. It takes a list
2701 of symlinks to create to file system sockets or FIFOs
2702 created by the specific Unix sockets. This is useful to
2703 manage symlinks to socket nodes with the same life-cycle as
2706 * The /dev/log socket and /dev/initctl FIFO have been moved to
2707 /run, and have been replaced by symlinks. This allows
2708 connecting to these facilities even if PrivateDevices=yes is
2709 used for a service (which makes /dev/log itself unavailable,
2710 but /run is left). This also has the benefit of ensuring
2711 that /dev only contains device nodes, directories and
2712 symlinks, and nothing else.
2714 * sd-daemon gained two new calls sd_pid_notify() and
2715 sd_pid_notifyf(). They are similar to sd_notify() and
2716 sd_notifyf(), but allow overriding of the source PID of
2717 notification messages if permissions permit this. This is
2718 useful to send notify messages on behalf of a different
2719 process (for example, the parent process). The
2720 systemd-notify tool has been updated to make use of this
2721 when sending messages (so that notification messages now
2722 originate from the shell script invoking systemd-notify and
2723 not the systemd-notify process itself. This should minimize
2724 a race where systemd fails to associate notification
2725 messages to services when the originating process already
2728 * A new "on-abnormal" setting for Restart= has been added. If
2729 set, it will result in automatic restarts on all "abnormal"
2730 reasons for a process to exit, which includes unclean
2731 signals, core dumps, timeouts and watchdog timeouts, but
2732 does not include clean and unclean exit codes or clean
2733 signals. Restart=on-abnormal is an alternative for
2734 Restart=on-failure for services that shall be able to
2735 terminate and avoid restarts on certain errors, by
2736 indicating so with an unclean exit code. Restart=on-failure
2737 or Restart=on-abnormal is now the recommended setting for
2738 all long-running services.
2740 * If the InaccessibleDirectories= service setting points to a
2741 mount point (or if there are any submounts contained within
2742 it), it is now attempted to completely unmount it, to make
2743 the file systems truly unavailable for the respective
2746 * The ReadOnlyDirectories= service setting and
2747 systemd-nspawn's --read-only parameter are now recursively
2748 applied to all submounts, too.
2750 * Mount units may now be created transiently via the bus APIs.
2752 * The support for SysV and LSB init scripts has been removed
2753 from the systemd daemon itself. Instead, it is now
2754 implemented as a generator that creates native systemd units
2755 from these scripts when needed. This enables us to remove a
2756 substantial amount of legacy code from PID 1, following the
2757 fact that many distributions only ship a very small number
2758 of LSB/SysV init scripts nowadays.
2760 * Privileged Xen (dom0) domains are not considered
2761 virtualization anymore by the virtualization detection
2762 logic. After all, they generally have unrestricted access to
2763 the hardware and usually are used to manage the unprivileged
2766 * systemd-tmpfiles gained a new "C" line type, for copying
2767 files or entire directories.
2769 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2770 lines. So far, they have been non-globbing versions of the
2771 latter, and have thus been redundant. In future, it is
2772 recommended to only use "z". "m" has hence been removed
2773 from the documentation, even though it stays supported.
2775 * A tmpfiles snippet to recreate the most basic structure in
2776 /var has been added. This is enough to create the /var/run →
2777 /run symlink and create a couple of structural
2778 directories. This allows systems to boot up with an empty or
2779 volatile /var. Of course, while with this change, the core OS
2780 now is capable with dealing with a volatile /var, not all
2781 user services are ready for it. However, we hope that sooner
2782 or later, many service daemons will be changed upstream so
2783 that they are able to automatically create their necessary
2784 directories in /var at boot, should they be missing. This is
2785 the first step to allow state-less systems that only require
2786 the vendor image for /usr to boot.
2788 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2789 empty tmpfs instance to a specific directory. This is
2790 particularly useful for making use of the automatic
2791 reconstruction of /var (see above), by passing --tmpfs=/var.
2793 * Access modes specified in tmpfiles snippets may now be
2794 prefixed with "~", which indicates that they shall be masked
2795 by whether the existing file or directory is currently
2796 writable, readable or executable at all. Also, if specified,
2797 the sgid/suid/sticky bits will be masked for all
2800 * A new passive target unit "network-pre.target" has been
2801 added which is useful for services that shall run before any
2802 network is configured, for example firewall scripts.
2804 * The "floppy" group that previously owned the /dev/fd*
2805 devices is no longer used. The "disk" group is now used
2806 instead. Distributions should probably deprecate usage of
2809 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2810 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2811 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2812 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2813 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2814 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2817 — Berlin, 2014-06-11
2821 * A new "systemd-timesyncd" daemon has been added for
2822 synchronizing the system clock across the network. It
2823 implements an SNTP client. In contrast to NTP
2824 implementations such as chrony or the NTP reference server,
2825 this only implements a client side, and does not bother with
2826 the full NTP complexity, focusing only on querying time from
2827 one remote server and synchronizing the local clock to
2828 it. Unless you intend to serve NTP to networked clients or
2829 want to connect to local hardware clocks, this simple NTP
2830 client should be more than appropriate for most
2831 installations. The daemon runs with minimal privileges, and
2832 has been hooked up with networkd to only operate when
2833 network connectivity is available. The daemon saves the
2834 current clock to disk every time a new NTP sync has been
2835 acquired, and uses this to possibly correct the system clock
2836 early at bootup, in order to accommodate for systems that
2837 lack an RTC such as the Raspberry Pi and embedded devices,
2838 and to make sure that time monotonically progresses on these
2839 systems, even if it is not always correct. To make use of
2840 this daemon, a new system user and group "systemd-timesync"
2841 needs to be created on installation of systemd.
2843 * The queue "seqnum" interface of libudev has been disabled, as
2844 it was generally incompatible with device namespacing as
2845 sequence numbers of devices go "missing" if the devices are
2846 part of a different namespace.
2848 * "systemctl list-timers" and "systemctl list-sockets" gained
2849 a --recursive switch for showing units of these types also
2850 for all local containers, similar in style to the already
2851 supported --recursive switch for "systemctl list-units".
2853 * A new RebootArgument= setting has been added for service
2854 units, which may be used to specify a kernel reboot argument
2855 to use when triggering reboots with StartLimitAction=.
2857 * A new FailureAction= setting has been added for service
2858 units which may be used to specify an operation to trigger
2859 when a service fails. This works similarly to
2860 StartLimitAction=, but unlike it, controls what is done
2861 immediately rather than only after several attempts to
2862 restart the service in question.
2864 * hostnamed got updated to also expose the kernel name,
2865 release, and version on the bus. This is useful for
2866 executing commands like hostnamectl with the -H switch.
2867 systemd-analyze makes use of this to properly display
2868 details when running non-locally.
2870 * The bootchart tool can now show cgroup information in the
2871 graphs it generates.
2873 * The CFS CPU quota cgroup attribute is now exposed for
2874 services. The new CPUQuota= switch has been added for this
2875 which takes a percentage value. Setting this will have the
2876 result that a service may never get more CPU time than the
2877 specified percentage, even if the machine is otherwise idle.
2879 * systemd-networkd learned IPIP and SIT tunnel support.
2881 * LSB init scripts exposing a dependency on $network will now
2882 get a dependency on network-online.target rather than simply
2883 network.target. This should bring LSB handling closer to
2884 what it was on SysV systems.
2886 * A new fsck.repair= kernel option has been added to control
2887 how fsck shall deal with unclean file systems at boot.
2889 * The (.ini) configuration file parser will now silently
2890 ignore sections whose name begins with "X-". This may be
2891 used to maintain application-specific extension sections in unit
2894 * machined gained a new API to query the IP addresses of
2895 registered containers. "machinectl status" has been updated
2896 to show these addresses in its output.
2898 * A new call sd_uid_get_display() has been added to the
2899 sd-login APIs for querying the "primary" session of a
2900 user. The "primary" session of the user is elected from the
2901 user's sessions and generally a graphical session is
2902 preferred over a text one.
2904 * A minimal systemd-resolved daemon has been added. It
2905 currently simply acts as a companion to systemd-networkd and
2906 manages resolv.conf based on per-interface DNS
2907 configuration, possibly supplied via DHCP. In the long run
2908 we hope to extend this into a local DNSSEC enabled DNS and
2911 * The systemd-networkd-wait-online tool is now enabled by
2912 default. It will delay network-online.target until a network
2913 connection has been configured. The tool primarily integrates
2914 with networkd, but will also make a best effort to make sense
2915 of network configuration performed in some other way.
2917 * Two new service options StartupCPUShares= and
2918 StartupBlockIOWeight= have been added that work similarly to
2919 CPUShares= and BlockIOWeight= however only apply during
2920 system startup. This is useful to prioritize certain services
2921 differently during bootup than during normal runtime.
2923 * hostnamed has been changed to prefer the statically
2924 configured hostname in /etc/hostname (unless set to
2925 'localhost' or empty) over any dynamic one supplied by
2926 dhcp. With this change, the rules for picking the hostname
2927 match more closely the rules of other configuration settings
2928 where the local administrator's configuration in /etc always
2929 overrides any other settings.
2931 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2932 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2933 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2934 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2935 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2936 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2937 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2938 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2939 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2940 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2941 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2942 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2943 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2944 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2945 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2946 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2949 — Beijing, 2014-05-28
2953 * When restoring the screen brightness at boot, stay away from
2954 the darkest setting or from the lowest 5% of the available
2955 range, depending on which is the larger value of both. This
2956 should effectively protect the user from rebooting into a
2957 black screen, should the brightness have been set to minimum
2960 * sd-login gained a new sd_machine_get_class() call to
2961 determine the class ("vm" or "container") of a machine
2962 registered with machined.
2964 * sd-login gained new calls
2965 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2966 to query the identity of the peer of a local AF_UNIX
2967 connection. They operate similarly to their sd_pid_get_xyz()
2970 * PID 1 will now maintain a system-wide system state engine
2971 with the states "starting", "running", "degraded",
2972 "maintenance", "stopping". These states are bound to system
2973 startup, normal runtime, runtime with at least one failed
2974 service, rescue/emergency mode and system shutdown. This
2975 state is shown in the "systemctl status" output when no unit
2976 name is passed. It is useful to determine system state, in
2977 particularly when doing so for many systems or containers at
2980 * A new command "list-machines" has been added to "systemctl"
2981 that lists all local OS containers and shows their system
2982 state (see above), if systemd runs inside of them.
2984 * systemctl gained a new "-r" switch to recursively enumerate
2985 units on all local containers, when used with the
2986 "list-unit" command (which is the default one that is
2987 executed when no parameters are specified).
2989 * The GPT automatic partition discovery logic will now honour
2990 two GPT partition flags: one may be set on a partition to
2991 cause it to be mounted read-only, and the other may be set
2992 on a partition to ignore it during automatic discovery.
2994 * Two new GPT type UUIDs have been added for automatic root
2995 partition discovery, for 32-bit and 64-bit ARM. This is not
2996 particularly useful for discovering the root directory on
2997 these architectures during bare-metal boots (since UEFI is
2998 not common there), but still very useful to allow booting of
2999 ARM disk images in nspawn with the -i option.
3001 * MAC addresses of interfaces created with nspawn's
3002 --network-interface= switch will now be generated from the
3003 machine name, and thus be stable between multiple invocations
3006 * logind will now automatically remove all IPC objects owned
3007 by a user if she or he fully logs out. This makes sure that
3008 users who are logged out cannot continue to consume IPC
3009 resources. This covers SysV memory, semaphores and message
3010 queues as well as POSIX shared memory and message
3011 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3012 limits. With this functionality, that is corrected. This may
3013 be turned off by using the RemoveIPC= switch of logind.conf.
3015 * The systemd-machine-id-setup and tmpfiles tools gained a
3016 --root= switch to operate on a specific root directory,
3019 * journald can now forward logged messages to the TTYs of all
3020 logged in users ("wall"). This is the default for all
3021 emergency messages now.
3023 * A new tool systemd-journal-remote has been added to stream
3024 journal log messages across the network.
3026 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3027 controller trees are mounted into it. Note that the
3028 directories mounted beneath it are not read-only. This is a
3029 security measure and is particularly useful because glibc
3030 actually includes a search logic to pick any tmpfs it can
3031 find to implement shm_open() if /dev/shm is not available
3032 (which it might very well be in namespaced setups).
3034 * machinectl gained a new "poweroff" command to cleanly power
3035 down a local OS container.
3037 * The PrivateDevices= unit file setting will now also drop the
3038 CAP_MKNOD capability from the capability bound set, and
3039 imply DevicePolicy=closed.
3041 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3042 comprehensively on all long-running systemd services where
3043 this is appropriate.
3045 * systemd-udevd will now run in a disassociated mount
3046 namespace. To mount directories from udev rules, make sure to
3047 pull in mount units via SYSTEMD_WANTS properties.
3049 * The kdbus support gained support for uploading policy into
3050 the kernel. sd-bus gained support for creating "monitoring"
3051 connections that can eavesdrop into all bus communication
3052 for debugging purposes.
3054 * Timestamps may now be specified in seconds since the UNIX
3055 epoch Jan 1st, 1970 by specifying "@" followed by the value
3058 * Native tcpwrap support in systemd has been removed. tcpwrap
3059 is old code, not really maintained anymore and has serious
3060 shortcomings, and better options such as firewalls
3061 exist. For setups that require tcpwrap usage, please
3062 consider invoking your socket-activated service via tcpd,
3063 like on traditional inetd.
3065 * A new system.conf configuration option
3066 DefaultTimerAccuracySec= has been added that controls the
3067 default AccuracySec= setting of .timer units.
3069 * Timer units gained a new WakeSystem= switch. If enabled,
3070 timers configured this way will cause the system to resume
3071 from system suspend (if the system supports that, which most
3074 * Timer units gained a new Persistent= switch. If enabled,
3075 timers configured this way will save to disk when they have
3076 been last triggered. This information is then used on next
3077 reboot to possible execute overdue timer events, that
3078 could not take place because the system was powered off.
3079 This enables simple anacron-like behaviour for timer units.
3081 * systemctl's "list-timers" will now also list the time a
3082 timer unit was last triggered in addition to the next time
3083 it will be triggered.
3085 * systemd-networkd will now assign predictable IPv4LL
3086 addresses to its local interfaces.
3088 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3089 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3090 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3091 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3092 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3093 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3094 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3095 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3098 — Berlin, 2014-03-25
3102 * A new unit file setting RestrictAddressFamilies= has been
3103 added to restrict which socket address families unit
3104 processes gain access to. This takes address family names
3105 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3106 attack surface of services via exotic protocol stacks. This
3107 is built on seccomp system call filters.
3109 * Two new unit file settings RuntimeDirectory= and
3110 RuntimeDirectoryMode= have been added that may be used to
3111 manage a per-daemon runtime directories below /run. This is
3112 an alternative for setting up directory permissions with
3113 tmpfiles snippets, and has the advantage that the runtime
3114 directory's lifetime is bound to the daemon runtime and that
3115 the daemon starts up with an empty directory each time. This
3116 is particularly useful when writing services that drop
3117 privileges using the User= or Group= setting.
3119 * The DeviceAllow= unit setting now supports globbing for
3120 matching against device group names.
3122 * The systemd configuration file system.conf gained new
3123 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3124 DefaultMemoryAccounting= to globally turn on/off accounting
3125 for specific resources (cgroups) for all units. These
3126 settings may still be overridden individually in each unit
3129 * systemd-gpt-auto-generator is now able to discover /srv and
3130 root partitions in addition to /home and swap partitions. It
3131 also supports LUKS-encrypted partitions now. With this in
3132 place, automatic discovery of partitions to mount following
3133 the Discoverable Partitions Specification
3134 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3135 is now a lot more complete. This allows booting without
3136 /etc/fstab and without root= on the kernel command line on
3137 systems prepared appropriately.
3139 * systemd-nspawn gained a new --image= switch which allows
3140 booting up disk images and Linux installations on any block
3141 device that follow the Discoverable Partitions Specification
3142 (see above). This means that installations made with
3143 appropriately updated installers may now be started and
3144 deployed using container managers, completely
3145 unmodified. (We hope that libvirt-lxc will add support for
3146 this feature soon, too.)
3148 * systemd-nspawn gained a new --network-macvlan= setting to
3149 set up a private macvlan interface for the
3150 container. Similarly, systemd-networkd gained a new
3151 Kind=macvlan setting in .netdev files.
3153 * systemd-networkd now supports configuring local addresses
3156 * A new tool systemd-network-wait-online has been added to
3157 synchronously wait for network connectivity using
3160 * The sd-bus.h bus API gained a new sd_bus_track object for
3161 tracking the life-cycle of bus peers. Note that sd-bus.h is
3162 still not a public API though (unless you specify
3163 --enable-kdbus on the configure command line, which however
3164 voids your warranty and you get no API stability guarantee).
3166 * The $XDG_RUNTIME_DIR runtime directories for each user are
3167 now individual tmpfs instances, which has the benefit of
3168 introducing separate pools for each user, with individual
3169 size limits, and thus making sure that unprivileged clients
3170 can no longer negatively impact the system or other users by
3171 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3172 RuntimeDirectorySize= has been introduced that allows
3173 controlling the default size limit for all users. It
3174 defaults to 10% of the available physical memory. This is no
3175 replacement for quotas on tmpfs though (which the kernel
3176 still does not support), as /dev/shm and /tmp are still
3177 shared resources used by both the system and unprivileged
3180 * logind will now automatically turn off automatic suspending
3181 on laptop lid close when more than one display is
3182 connected. This was previously expected to be implemented
3183 individually in desktop environments (such as GNOME),
3184 however has been added to logind now, in order to fix a
3185 boot-time race where a desktop environment might not have
3186 been started yet and thus not been able to take an inhibitor
3187 lock at the time where logind already suspends the system
3188 due to a closed lid.
3190 * logind will now wait at least 30s after each system
3191 suspend/resume cycle, and 3min after system boot before
3192 suspending the system due to a closed laptop lid. This
3193 should give USB docking stations and similar enough time to
3194 be probed and configured after system resume and boot in
3195 order to then act as suspend blocker.
3197 * systemd-run gained a new --property= setting which allows
3198 initialization of resource control properties (and others)
3199 for the created scope or service unit. Example: "systemd-run
3200 --property=BlockIOWeight=10 updatedb" may be used to run
3201 updatedb at a low block IO scheduling weight.
3203 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3204 now also work in --scope mode.
3206 * When systemd is compiled with kdbus support, basic support
3207 for enforced policies is now in place. (Note that enabling
3208 kdbus still voids your warranty and no API compatibility
3211 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3212 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3213 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3214 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3215 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3216 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3217 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3218 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3219 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3220 Zbigniew Jędrzejewski-Szmek
3222 — Berlin, 2014-03-12
3226 * systemd will now relabel /dev after loading the SMACK policy
3227 according to SMACK rules.
3229 * A new unit file option AppArmorProfile= has been added to
3230 set the AppArmor profile for the processes of a unit.
3232 * A new condition check ConditionArchitecture= has been added
3233 to conditionalize units based on the system architecture, as
3234 reported by uname()'s "machine" field.
3236 * systemd-networkd now supports matching on the system
3237 virtualization, architecture, kernel command line, host name
3240 * logind is now a lot more aggressive when suspending the
3241 machine due to a closed laptop lid. Instead of acting only
3242 on the lid close action, it will continuously watch the lid
3243 status and act on it. This is useful for laptops where the
3244 power button is on the outside of the chassis so that it can
3245 be reached without opening the lid (such as the Lenovo
3246 Yoga). On those machines, logind will now immediately
3247 re-suspend the machine if the power button has been
3248 accidentally pressed while the laptop was suspended and in a
3249 backpack or similar.
3251 * logind will now watch SW_DOCK switches and inhibit reaction
3252 to the lid switch if it is pressed. This means that logind
3253 will not suspend the machine anymore if the lid is closed
3254 and the system is docked, if the laptop supports SW_DOCK
3255 notifications via the input layer. Note that ACPI docking
3256 stations do not generate this currently. Also note that this
3257 logic is usually not fully sufficient and Desktop
3258 Environments should take a lid switch inhibitor lock when an
3259 external display is connected, as systemd will not watch
3262 * nspawn will now make use of the devices cgroup controller by
3263 default, and only permit creation of and access to the usual
3264 API device nodes like /dev/null or /dev/random, as well as
3265 access to (but not creation of) the pty devices.
3267 * We will now ship a default .network file for
3268 systemd-networkd that automatically configures DHCP for
3269 network interfaces created by nspawn's --network-veth or
3270 --network-bridge= switches.
3272 * systemd will now understand the usual M, K, G, T suffixes
3273 according to SI conventions (i.e. to the base 1000) when
3274 referring to throughput and hardware metrics. It will stay
3275 with IEC conventions (i.e. to the base 1024) for software
3276 metrics, according to what is customary according to
3277 Wikipedia. We explicitly document which base applies for
3278 each configuration option.
3280 * The DeviceAllow= setting in unit files now supports a syntax
3281 to whitelist an entire group of devices node majors at once,
3282 based on the /proc/devices listing. For example, with the
3283 string "char-pts", it is now possible to whitelist all
3284 current and future pseudo-TTYs at once.
3286 * sd-event learned a new "post" event source. Event sources of
3287 this type are triggered by the dispatching of any event
3288 source of a type that is not "post". This is useful for
3289 implementing clean-up and check event sources that are
3290 triggered by other work being done in the program.
3292 * systemd-networkd is no longer statically enabled, but uses
3293 the usual [Install] sections so that it can be
3294 enabled/disabled using systemctl. It still is enabled by
3297 * When creating a veth interface pair with systemd-nspawn, the
3298 host side will now be prefixed with "vb-" if
3299 --network-bridge= is used, and with "ve-" if --network-veth
3300 is used. This way, it is easy to distinguish these cases on
3301 the host, for example to apply different configuration to
3302 them with systemd-networkd.
3304 * The compatibility libraries for libsystemd-journal.so,
3305 libsystem-id128.so, libsystemd-login.so and
3306 libsystemd-daemon.so do not make use of IFUNC
3307 anymore. Instead, we now build libsystemd.so multiple times
3308 under these alternative names. This means that the footprint
3309 is drastically increased, but given that these are
3310 transitional compatibility libraries, this should not matter
3311 much. This change has been made necessary to support the ARM
3312 platform for these compatibility libraries, as the ARM
3313 toolchain is not really at the same level as the toolchain
3314 for other architectures like x86 and does not support
3315 IFUNC. Please make sure to use --enable-compat-libs only
3316 during a transitional period!
3318 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3319 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3320 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3321 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3322 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3323 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3324 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3325 Zbigniew Jędrzejewski-Szmek
3327 — Berlin, 2014-02-24
3331 * A new component "systemd-networkd" has been added that can
3332 be used to configure local network interfaces statically or
3333 via DHCP. It is capable of bringing up bridges, VLANs, and
3334 bonding. Currently, no hook-ups for interactive network
3335 configuration are provided. Use this for your initrd,
3336 container, embedded, or server setup if you need a simple,
3337 yet powerful, network configuration solution. This
3338 configuration subsystem is quite nifty, as it allows wildcard
3339 hotplug matching in interfaces. For example, with a single
3340 configuration snippet, you can configure that all Ethernet
3341 interfaces showing up are automatically added to a bridge,
3342 or similar. It supports link-sensing and more.
3344 * A new tool "systemd-socket-proxyd" has been added which can
3345 act as a bidirectional proxy for TCP sockets. This is
3346 useful for adding socket activation support to services that
3347 do not actually support socket activation, including virtual
3348 machines and the like.
3350 * Add a new tool to save/restore rfkill state on
3353 * Save/restore state of keyboard backlights in addition to
3354 display backlights on shutdown/boot.
3356 * udev learned a new SECLABEL{} construct to label device
3357 nodes with a specific security label when they appear. For
3358 now, only SECLABEL{selinux} is supported, but the syntax is
3359 prepared for additional security frameworks.
3361 * udev gained a new scheme to configure link-level attributes
3362 from files in /etc/systemd/network/*.link. These files can
3363 match against MAC address, device path, driver name and type,
3364 and will apply attributes like the naming policy, link speed,
3365 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3366 address assignment policy (randomized, ...).
3368 * The configuration of network interface naming rules for
3369 "permanent interface names" has changed: a new NamePolicy=
3370 setting in the [Link] section of .link files determines the
3371 priority of possible naming schemes (onboard, slot, MAC,
3372 path). The default value of this setting is determined by
3373 /usr/lib/net/links/99-default.link. Old
3374 80-net-name-slot.rules udev configuration file has been
3375 removed, so local configuration overriding this file should
3376 be adapated to override 99-default.link instead.
3378 * When the User= switch is used in a unit file, also
3379 initialize $SHELL= based on the user database entry.
3381 * systemd no longer depends on libdbus. All communication is
3382 now done with sd-bus, systemd's low-level bus library
3385 * kdbus support has been added to PID 1 itself. When kdbus is
3386 enabled, this causes PID 1 to set up the system bus and
3387 enable support for a new ".busname" unit type that
3388 encapsulates bus name activation on kdbus. It works a little
3389 bit like ".socket" units, except for bus names. A new
3390 generator has been added that converts classic dbus1 service
3391 activation files automatically into native systemd .busname
3394 * sd-bus: add a light-weight vtable implementation that allows
3395 defining objects on the bus with a simple static const
3396 vtable array of its methods, signals and properties.
3398 * systemd will not generate or install static dbus
3399 introspection data anymore to /usr/share/dbus-1/interfaces,
3400 as the precise format of these files is unclear, and
3401 nothing makes use of it.
3403 * A proxy daemon is now provided to proxy clients connecting
3404 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3405 compatibility with classic D-Bus.
3407 * A bus driver implementation has been added that supports the
3408 classic D-Bus bus driver calls on kdbus, also for
3409 compatibility purposes.
3411 * A new API "sd-event.h" has been added that implements a
3412 minimal event loop API built around epoll. It provides a
3413 couple of features that direct epoll usage is lacking:
3414 prioritization of events, scales to large numbers of timer
3415 events, per-event timer slack (accuracy), system-wide
3416 coalescing of timer events, exit handlers, watchdog
3417 supervision support using systemd's sd_notify() API, child
3420 * A new API "sd-rntl.h" has been added that provides an API
3421 around the route netlink interface of the kernel, similar in
3422 style to "sd-bus.h".
3424 * A new API "sd-dhcp-client.h" has been added that provides a
3425 small DHCPv4 client-side implementation. This is used by
3428 * There is a new kernel command line option
3429 "systemd.restore_state=0|1". When set to "0", none of the
3430 systemd tools will restore saved runtime state to hardware
3431 devices. More specifically, the rfkill and backlight states
3434 * The FsckPassNo= compatibility option in mount/service units
3435 has been removed. The fstab generator will now add the
3436 necessary dependencies automatically, and does not require
3437 PID1's support for that anymore.
3439 * journalctl gained a new switch, --list-boots, that lists
3440 recent boots with their times and boot IDs.
3442 * The various tools like systemctl, loginctl, timedatectl,
3443 busctl, systemd-run, ... have gained a new switch "-M" to
3444 connect to a specific, local OS container (as direct
3445 connection, without requiring SSH). This works on any
3446 container that is registered with machined, such as those
3447 created by libvirt-lxc or nspawn.
3449 * systemd-run and systemd-analyze also gained support for "-H"
3450 to connect to remote hosts via SSH. This is particularly
3451 useful for systemd-run because it enables queuing of jobs
3452 onto remote systems.
3454 * machinectl gained a new command "login" to open a getty
3455 login in any local container. This works with any container
3456 that is registered with machined (such as those created by
3457 libvirt-lxc or nspawn), and which runs systemd inside.
3459 * machinectl gained a new "reboot" command that may be used to
3460 trigger a reboot on a specific container that is registered
3461 with machined. This works on any container that runs an init
3462 system of some kind.
3464 * systemctl gained a new "list-timers" command to print a nice
3465 listing of installed timer units with the times they elapse
3468 * Alternative reboot() parameters may now be specified on the
3469 "systemctl reboot" command line and are passed to the
3470 reboot() system call.
3472 * systemctl gained a new --job-mode= switch to configure the
3473 mode to queue a job with. This is a more generic version of
3474 --fail, --irreversible, and --ignore-dependencies, which are
3475 still available but not advertised anymore.
3477 * /etc/systemd/system.conf gained new settings to configure
3478 various default timeouts of units, as well as the default
3479 start limit interval and burst. These may still be overridden
3482 * PID1 will now export on the bus profile data of the security
3483 policy upload process (such as the SELinux policy upload to
3486 * journald: when forwarding logs to the console, include
3487 timestamps (following the setting in
3488 /sys/module/printk/parameters/time).
3490 * OnCalendar= in timer units now understands the special
3491 strings "yearly" and "annually". (Both are equivalent)
3493 * The accuracy of timer units is now configurable with the new
3494 AccuracySec= setting. It defaults to 1min.
3496 * A new dependency type JoinsNamespaceOf= has been added that
3497 allows running two services within the same /tmp and network
3498 namespace, if PrivateNetwork= or PrivateTmp= are used.
3500 * A new command "cat" has been added to systemctl. It outputs
3501 the original unit file of a unit, and concatenates the
3502 contents of additional "drop-in" unit file snippets, so that
3503 the full configuration is shown.
3505 * systemctl now supports globbing on the various "list-xyz"
3506 commands, like "list-units" or "list-sockets", as well as on
3507 those commands which take multiple unit names.
3509 * journalctl's --unit= switch gained support for globbing.
3511 * All systemd daemons now make use of the watchdog logic so
3512 that systemd automatically notices when they hang.
3514 * If the $container_ttys environment variable is set,
3515 getty-generator will automatically spawn a getty for each
3516 listed tty. This is useful for container managers to request
3517 login gettys to be spawned on as many ttys as needed.
3519 * %h, %s, %U specifier support is not available anymore when
3520 used in unit files for PID 1. This is because NSS calls are
3521 not safe from PID 1. They stay available for --user
3522 instances of systemd, and as special case for the root user.
3524 * loginctl gained a new "--no-legend" switch to turn off output
3527 * The "sd-login.h" API gained three new calls:
3528 sd_session_is_remote(), sd_session_get_remote_user(),
3529 sd_session_get_remote_host() to query information about
3532 * The udev hardware database now also carries vendor/product
3533 information of SDIO devices.
3535 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3536 determine whether watchdog notifications are requested by
3539 * Socket-activated per-connection services now include a
3540 short description of the connection parameters in the
3543 * tmpfiles gained a new "--boot" option. When this is not used,
3544 only lines where the command character is not suffixed with
3545 "!" are executed. When this option is specified, those
3546 options are executed too. This partitions tmpfiles
3547 directives into those that can be safely executed at any
3548 time, and those which should be run only at boot (for
3549 example, a line that creates /run/nologin).
3551 * A new API "sd-resolve.h" has been added which provides a simple
3552 asynchronous wrapper around glibc NSS host name resolution
3553 calls, such as getaddrinfo(). In contrast to glibc's
3554 getaddrinfo_a(), it does not use signals. In contrast to most
3555 other asynchronous name resolution libraries, this one does
3556 not reimplement DNS, but reuses NSS, so that alternate
3557 host name resolution systems continue to work, such as mDNS,
3558 LDAP, etc. This API is based on libasyncns, but it has been
3559 cleaned up for inclusion in systemd.
3561 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3562 "sd-daemon.h" are no longer found in individual libraries
3563 libsystemd-journal.so, libsystemd-login.so,
3564 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3565 merged them into a single library, libsystemd.so, which
3566 provides all symbols. The reason for this is cyclic
3567 dependencies, as these libraries tend to use each other's
3568 symbols. So far, we have managed to workaround that by linking
3569 a copy of a good part of our code into each of these
3570 libraries again and again, which, however, makes certain
3571 things hard to do, like sharing static variables. Also, it
3572 substantially increases footprint. With this change, there
3573 is only one library for the basic APIs systemd
3574 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3575 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3576 library as well, however are subject to the --enable-kdbus
3577 switch (see below). Note that "sd-dhcp-client.h" is not part
3578 of this library (this is because it only consumes, never
3579 provides, services of/to other APIs). To make the transition
3580 easy from the separate libraries to the unified one, we
3581 provide the --enable-compat-libs compile-time switch which
3582 will generate stub libraries that are compatible with the
3583 old ones but redirect all calls to the new one.
3585 * All of the kdbus logic and the new APIs "sd-bus.h",
3586 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3587 and "sd-utf8.h" are compile-time optional via the
3588 "--enable-kdbus" switch, and they are not compiled in by
3589 default. To make use of kdbus, you have to explicitly enable
3590 the switch. Note however, that neither the kernel nor the
3591 userspace API for all of this is considered stable yet. We
3592 want to maintain the freedom to still change the APIs for
3593 now. By specifying this build-time switch, you acknowledge
3594 that you are aware of the instability of the current
3597 * Also, note that while kdbus is pretty much complete,
3598 it lacks one thing: proper policy support. This means you
3599 can build a fully working system with all features; however,
3600 it will be highly insecure. Policy support will be added in
3601 one of the next releases, at the same time that we will
3602 declare the APIs stable.
3604 * When the kernel command line argument "kdbus" is specified,
3605 systemd will automatically load the kdbus.ko kernel module. At
3606 this stage of development, it is only useful for testing kdbus
3607 and should not be used in production. Note: if "--enable-kdbus"
3608 is specified, and the kdbus.ko kernel module is available, and
3609 "kdbus" is added to the kernel command line, the entire system
3610 runs with kdbus instead of dbus-daemon, with the above mentioned
3611 problem of missing the system policy enforcement. Also a future
3612 version of kdbus.ko or a newer systemd will not be compatible with
3613 each other, and will unlikely be able to boot the machine if only
3614 one of them is updated.
3616 * systemctl gained a new "import-environment" command which
3617 uploads the caller's environment (or parts thereof) into the
3618 service manager so that it is inherited by services started
3619 by the manager. This is useful to upload variables like
3620 $DISPLAY into the user service manager.
3622 * A new PrivateDevices= switch has been added to service units
3623 which allows running a service with a namespaced /dev
3624 directory that does not contain any device nodes for
3625 physical devices. More specifically, it only includes devices
3626 such as /dev/null, /dev/urandom, and /dev/zero which are API
3629 * logind has been extended to support behaviour like VT
3630 switching on seats that do not support a VT. This makes
3631 multi-session available on seats that are not the first seat
3632 (seat0), and on systems where kernel support for VTs has
3633 been disabled at compile-time.
3635 * If a process holds a delay lock for system sleep or shutdown
3636 and fails to release it in time, we will now log its
3637 identity. This makes it easier to identify processes that
3638 cause slow suspends or power-offs.
3640 * When parsing /etc/crypttab, support for a new key-slot=
3641 option as supported by Debian is added. It allows indicating
3642 which LUKS slot to use on disk, speeding up key loading.
3644 * The sd_journald_sendv() API call has been checked and
3645 officially declared to be async-signal-safe so that it may
3646 be invoked from signal handlers for logging purposes.
3648 * Boot-time status output is now enabled automatically after a
3649 short timeout if boot does not progress, in order to give
3650 the user an indication what she or he is waiting for.
3652 * The boot-time output has been improved to show how much time
3653 remains until jobs expire.
3655 * The KillMode= switch in service units gained a new possible
3656 value "mixed". If set, and the unit is shut down, then the
3657 initial SIGTERM signal is sent only to the main daemon
3658 process, while the following SIGKILL signal is sent to
3659 all remaining processes of the service.
3661 * When a scope unit is registered, a new property "Controller"
3662 may be set. If set to a valid bus name, systemd will send a
3663 RequestStop() signal to this name when it would like to shut
3664 down the scope. This may be used to hook manager logic into
3665 the shutdown logic of scope units. Also, scope units may now
3666 be put in a special "abandoned" state, in which case the
3667 manager process which created them takes no further
3668 responsibilities for it.
3670 * When reading unit files, systemd will now verify
3671 the access mode of these files, and warn about certain
3672 suspicious combinations. This has been added to make it
3673 easier to track down packaging bugs where unit files are
3674 marked executable or world-writable.
3676 * systemd-nspawn gained a new "--setenv=" switch to set
3677 container-wide environment variables. The similar option in
3678 systemd-activate was renamed from "--environment=" to
3679 "--setenv=" for consistency.
3681 * systemd-nspawn has been updated to create a new kdbus domain
3682 for each container that is invoked, thus allowing each
3683 container to have its own set of system and user buses,
3684 independent of the host.
3686 * systemd-nspawn gained a new --drop-capability= switch to run
3687 the container with less capabilities than the default. Both
3688 --drop-capability= and --capability= now take the special
3689 string "all" for dropping or keeping all capabilities.
3691 * systemd-nspawn gained new switches for executing containers
3692 with specific SELinux labels set.
3694 * systemd-nspawn gained a new --quiet switch to not generate
3695 any additional output but the container's own console
3698 * systemd-nspawn gained a new --share-system switch to run a
3699 container without PID namespacing enabled.
3701 * systemd-nspawn gained a new --register= switch to control
3702 whether the container is registered with systemd-machined or
3703 not. This is useful for containers that do not run full
3704 OS images, but only specific apps.
3706 * systemd-nspawn gained a new --keep-unit which may be used
3707 when invoked as the only program from a service unit, and
3708 results in registration of the unit service itself in
3709 systemd-machined, instead of a newly opened scope unit.
3711 * systemd-nspawn gained a new --network-interface= switch for
3712 moving arbitrary interfaces to the container. The new
3713 --network-veth switch creates a virtual Ethernet connection
3714 between host and container. The new --network-bridge=
3715 switch then allows assigning the host side of this virtual
3716 Ethernet connection to a bridge device.
3718 * systemd-nspawn gained a new --personality= switch for
3719 setting the kernel personality for the container. This is
3720 useful when running a 32-bit container on a 64-bit host. A
3721 similar option Personality= is now also available for service
3724 * logind will now also track a "Desktop" identifier for each
3725 session which encodes the desktop environment of it. This is
3726 useful for desktop environments that want to identify
3727 multiple running sessions of itself easily.
3729 * A new SELinuxContext= setting for service units has been
3730 added that allows setting a specific SELinux execution
3731 context for a service.
3733 * Most systemd client tools will now honour $SYSTEMD_LESS for
3734 settings of the "less" pager. By default, these tools will
3735 override $LESS to allow certain operations to work, such as
3736 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3737 influence this logic.
3739 * systemd's "seccomp" hook-up has been changed to make use of
3740 the libseccomp library instead of using its own
3741 implementation. This has benefits for portability among
3744 * For usage together with SystemCallFilter=, a new
3745 SystemCallErrorNumber= setting has been introduced that
3746 allows configuration of a system error number to be returned
3747 on filtered system calls, instead of immediately killing the
3748 process. Also, SystemCallArchitectures= has been added to
3749 limit access to system calls of a particular architecture
3750 (in order to turn off support for unused secondary
3751 architectures). There is also a global
3752 SystemCallArchitectures= setting in system.conf now to turn
3753 off support for non-native system calls system-wide.
3755 * systemd requires a kernel with a working name_to_handle_at(),
3756 please see the kernel config requirements in the README file.
3758 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3759 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3760 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3761 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3762 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3763 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3764 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3765 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3766 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3767 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3768 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3769 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3770 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3771 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3772 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3773 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3774 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3775 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3776 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3777 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3778 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3779 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3780 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3781 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3783 — Berlin, 2014-02-20
3787 * logind has gained support for facilitating privileged input
3788 and drm device access for unprivileged clients. This work is
3789 useful to allow Wayland display servers (and similar
3790 programs, such as kmscon) to run under the user's ID and
3791 access input and drm devices which are normally
3792 protected. When this is used (and the kernel is new enough)
3793 logind will "mute" IO on the file descriptors passed to
3794 Wayland as long as it is in the background and "unmute" it
3795 if it returns into the foreground. This allows secure
3796 session switching without allowing background sessions to
3797 eavesdrop on input and display data. This also introduces
3798 session switching support if VT support is turned off in the
3799 kernel, and on seats that are not seat0.
3801 * A new kernel command line option luks.options= is understood
3802 now which allows specifying LUKS options for usage for LUKS
3803 encrypted partitions specified with luks.uuid=.
3805 * tmpfiles.d(5) snippets may now use specifier expansion in
3806 path names. More specifically %m, %b, %H, %v, are now
3807 replaced by the local machine id, boot id, hostname, and
3808 kernel version number.
3810 * A new tmpfiles.d(5) command "m" has been introduced which
3811 may be used to change the owner/group/access mode of a file
3812 or directory if it exists, but do nothing if it does not.
3814 * This release removes high-level support for the
3815 MemorySoftLimit= cgroup setting. The underlying kernel
3816 cgroup attribute memory.soft_limit= is currently badly
3817 designed and likely to be removed from the kernel API in its
3818 current form, hence we should not expose it for now.
3820 * The memory.use_hierarchy cgroup attribute is now enabled for
3821 all cgroups systemd creates in the memory cgroup
3822 hierarchy. This option is likely to be come the built-in
3823 default in the kernel anyway, and the non-hierarchical mode
3824 never made much sense in the intrinsically hierarchical
3827 * A new field _SYSTEMD_SLICE= is logged along with all journal
3828 messages containing the slice a message was generated
3829 from. This is useful to allow easy per-customer filtering of
3830 logs among other things.
3832 * systemd-journald will no longer adjust the group of journal
3833 files it creates to the "systemd-journal" group. Instead we
3834 rely on the journal directory to be owned by the
3835 "systemd-journal" group, and its setgid bit set, so that the
3836 kernel file system layer will automatically enforce that
3837 journal files inherit this group assignment. The reason for
3838 this change is that we cannot allow NSS look-ups from
3839 journald which would be necessary to resolve
3840 "systemd-journal" to a numeric GID, because this might
3841 create deadlocks if NSS involves synchronous queries to
3842 other daemons (such as nscd, or sssd) which in turn are
3843 logging clients of journald and might block on it, which
3844 would then dead lock. A tmpfiles.d(5) snippet included in
3845 systemd will make sure the setgid bit and group are
3846 properly set on the journal directory if it exists on every
3847 boot. However, we recommend adjusting it manually after
3848 upgrades too (or from RPM scriptlets), so that the change is
3849 not delayed until next reboot.
3851 * Backlight and random seed files in /var/lib/ have moved into
3852 the /var/lib/systemd/ directory, in order to centralize all
3853 systemd generated files in one directory.
3855 * Boot time performance measurements (as displayed by
3856 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3857 performance information if that's available to determine how
3858 much time BIOS and boot loader initialization required. With
3859 a sufficiently new BIOS you hence no longer need to boot
3860 with Gummiboot to get access to such information.
3862 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3863 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3864 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3865 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3866 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3867 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3868 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3870 — Berlin, 2013-10-02
3874 * The Restart= option for services now understands a new
3875 on-watchdog setting, which will restart the service
3876 automatically if the service stops sending out watchdog keep
3877 alive messages (as configured with WatchdogSec=).
3879 * The getty generator (which is responsible for bringing up a
3880 getty on configured serial consoles) will no longer only
3881 start a getty on the primary kernel console but on all
3882 others, too. This makes the order in which console= is
3883 specified on the kernel command line less important.
3885 * libsystemd-logind gained a new sd_session_get_vt() call to
3886 retrieve the VT number of a session.
3888 * If the option "tries=0" is set for an entry of /etc/crypttab
3889 its passphrase is queried indefinitely instead of any
3890 maximum number of tries.
3892 * If a service with a configure PID file terminates its PID
3893 file will now be removed automatically if it still exists
3894 afterwards. This should put an end to stale PID files.
3896 * systemd-run will now also take relative binary path names
3897 for execution and no longer insists on absolute paths.
3899 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3900 paths that are optionally prefixed with "-" to indicate that
3901 it should not be considered a failure if they do not exist.
3903 * journalctl -o (and similar commands) now understands a new
3904 output mode "short-precise", it is similar to "short" but
3905 shows timestamps with usec accuracy.
3907 * The option "discard" (as known from Debian) is now
3908 synonymous to "allow-discards" in /etc/crypttab. In fact,
3909 "discard" is preferred now (since it is easier to remember
3912 * Some licensing clean-ups were made, so that more code is now
3913 LGPL-2.1 licensed than before.
3915 * A minimal tool to save/restore the display backlight
3916 brightness across reboots has been added. It will store the
3917 backlight setting as late as possible at shutdown, and
3918 restore it as early as possible during reboot.
3920 * A logic to automatically discover and enable home and swap
3921 partitions on GPT disks has been added. With this in place
3922 /etc/fstab becomes optional for many setups as systemd can
3923 discover certain partitions located on the root disk
3924 automatically. Home partitions are recognized under their
3925 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3926 partitions are recognized under their GPT type ID
3927 0657fd6da4ab43c484e50933c84b4f4f.
3929 * systemd will no longer pass any environment from the kernel
3930 or initrd to system services. If you want to set an
3931 environment for all services, do so via the kernel command
3932 line systemd.setenv= assignment.
3934 * The systemd-sysctl tool no longer natively reads the file
3935 /etc/sysctl.conf. If desired, the file should be symlinked
3936 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3937 legacy support by a symlink rather than built-in code, it
3938 also makes the otherwise hidden order of application of the
3939 different files visible. (Note that this partly reverts to a
3940 pre-198 application order of sysctl knobs!)
3942 * The "systemctl set-log-level" and "systemctl dump" commands
3943 have been moved to systemd-analyze.
3945 * systemd-run learned the new --remain-after-exit switch,
3946 which causes the scope unit not to be cleaned up
3947 automatically after the process terminated.
3949 * tmpfiles learned a new --exclude-prefix= switch to exclude
3950 certain paths from operation.
3952 * journald will now automatically flush all messages to disk
3953 as soon as a message at the log level CRIT, ALERT or EMERG
3956 Contributions from: Andrew Cook, Brandon Philips, Christian
3957 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3958 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3959 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3960 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3961 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3962 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3963 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3964 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3965 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3966 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3967 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3968 William Giokas, Zbigniew Jędrzejewski-Szmek
3970 — Berlin, 2013-09-13
3974 * The documentation has been updated to cover the various new
3975 concepts introduced with 205.
3977 * Unit files now understand the new %v specifier which
3978 resolves to the kernel version string as returned by "uname
3981 * systemctl now supports filtering the unit list output by
3982 load state, active state and sub state, using the new
3985 * "systemctl status" will now show the results of the
3986 condition checks (like ConditionPathExists= and similar) of
3987 the last start attempts of the unit. They are also logged to
3990 * "journalctl -b" may now be used to look for boot output of a
3991 specific boot. Try "journalctl -b -1" for the previous boot,
3992 but the syntax is substantially more powerful.
3994 * "journalctl --show-cursor" has been added which prints the
3995 cursor string the last shown log line. This may then be used
3996 with the new "journalctl --after-cursor=" switch to continue
3997 browsing logs from that point on.
3999 * "journalctl --force" may now be used to force regeneration
4002 * Creation of "dead" device nodes has been moved from udev
4003 into kmod and tmpfiles. Previously, udev would read the kmod
4004 databases to pre-generate dead device nodes based on meta
4005 information contained in kernel modules, so that these would
4006 be auto-loaded on access rather then at boot. As this
4007 does not really have much to do with the exposing actual
4008 kernel devices to userspace this has always been slightly
4009 alien in the udev codebase. Following the new scheme kmod
4010 will now generate a runtime snippet for tmpfiles from the
4011 module meta information and it now is tmpfiles' job to the
4012 create the nodes. This also allows overriding access and
4013 other parameters for the nodes using the usual tmpfiles
4014 facilities. As side effect this allows us to remove the
4015 CAP_SYS_MKNOD capability bit from udevd entirely.
4017 * logind's device ACLs may now be applied to these "dead"
4018 devices nodes too, thus finally allowing managed access to
4019 devices such as /dev/snd/sequencer whithout loading the
4020 backing module right-away.
4022 * A new RPM macro has been added that may be used to apply
4023 tmpfiles configuration during package installation.
4025 * systemd-detect-virt and ConditionVirtualization= now can
4026 detect User-Mode-Linux machines (UML).
4028 * journald will now implicitly log the effective capabilities
4029 set of processes in the message metadata.
4031 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4033 * The initrd interface has been simplified (more specifically,
4034 support for passing performance data via environment
4035 variables and fsck results via files in /run has been
4036 removed). These features were non-essential, and are
4037 nowadays available in a much nicer way by having systemd in
4038 the initrd serialize its state and have the hosts systemd
4039 deserialize it again.
4041 * The udev "keymap" data files and tools to apply keyboard
4042 specific mappings of scan to key codes, and force-release
4043 scan code lists have been entirely replaced by a udev
4044 "keyboard" builtin and a hwdb data file.
4046 * systemd will now honour the kernel's "quiet" command line
4047 argument also during late shutdown, resulting in a
4048 completely silent shutdown when used.
4050 * There's now an option to control the SO_REUSEPORT socket
4051 option in .socket units.
4053 * Instance units will now automatically get a per-template
4054 subslice of system.slice unless something else is explicitly
4055 configured. For example, instances of sshd@.service will now
4056 implicitly be placed in system-sshd.slice rather than
4057 system.slice as before.
4059 * Test coverage support may now be enabled at build time.
4061 Contributions from: Dave Reisner, Frederic Crozat, Harald
4062 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4063 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4064 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4065 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4066 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4067 Giokas, Zbigniew Jędrzejewski-Szmek
4069 — Berlin, 2013-07-23
4073 * Two new unit types have been introduced:
4075 Scope units are very similar to service units, however, are
4076 created out of pre-existing processes — instead of PID 1
4077 forking off the processes. By using scope units it is
4078 possible for system services and applications to group their
4079 own child processes (worker processes) in a powerful way
4080 which then maybe used to organize them, or kill them
4081 together, or apply resource limits on them.
4083 Slice units may be used to partition system resources in an
4084 hierarchical fashion and then assign other units to them. By
4085 default there are now three slices: system.slice (for all
4086 system services), user.slice (for all user sessions),
4087 machine.slice (for VMs and containers).
4089 Slices and scopes have been introduced primarily in
4090 context of the work to move cgroup handling to a
4091 single-writer scheme, where only PID 1
4092 creates/removes/manages cgroups.
4094 * There's a new concept of "transient" units. In contrast to
4095 normal units these units are created via an API at runtime,
4096 not from configuration from disk. More specifically this
4097 means it is now possible to run arbitrary programs as
4098 independent services, with all execution parameters passed
4099 in via bus APIs rather than read from disk. Transient units
4100 make systemd substantially more dynamic then it ever was,
4101 and useful as a general batch manager.
4103 * logind has been updated to make use of scope and slice units
4104 for managing user sessions. As a user logs in he will get
4105 his own private slice unit, to which all sessions are added
4106 as scope units. We also added support for automatically
4107 adding an instance of user@.service for the user into the
4108 slice. Effectively logind will no longer create cgroup
4109 hierarchies on its own now, it will defer entirely to PID 1
4110 for this by means of scope, service and slice units. Since
4111 user sessions this way become entities managed by PID 1
4112 the output of "systemctl" is now a lot more comprehensive.
4114 * A new mini-daemon "systemd-machined" has been added which
4115 may be used by virtualization managers to register local
4116 VMs/containers. nspawn has been updated accordingly, and
4117 libvirt will be updated shortly. machined will collect a bit
4118 of meta information about the VMs/containers, and assign
4119 them their own scope unit (see above). The collected
4120 meta-data is then made available via the "machinectl" tool,
4121 and exposed in "ps" and similar tools. machined/machinectl
4122 is compile-time optional.
4124 * As discussed earlier, the low-level cgroup configuration
4125 options ControlGroup=, ControlGroupModify=,
4126 ControlGroupPersistent=, ControlGroupAttribute= have been
4127 removed. Please use high-level attribute settings instead as
4128 well as slice units.
4130 * A new bus call SetUnitProperties() has been added to alter
4131 various runtime parameters of a unit. This is primarily
4132 useful to alter cgroup parameters dynamically in a nice way,
4133 but will be extended later on to make more properties
4134 modifiable at runtime. systemctl gained a new set-properties
4135 command that wraps this call.
4137 * A new tool "systemd-run" has been added which can be used to
4138 run arbitrary command lines as transient services or scopes,
4139 while configuring a number of settings via the command
4140 line. This tool is currently very basic, however already
4141 very useful. We plan to extend this tool to even allow
4142 queuing of execution jobs with time triggers from the
4143 command line, similar in fashion to "at".
4145 * nspawn will now inform the user explicitly that kernels with
4146 audit enabled break containers, and suggest the user to turn
4149 * Support for detecting the IMA and AppArmor security
4150 frameworks with ConditionSecurity= has been added.
4152 * journalctl gained a new "-k" switch for showing only kernel
4153 messages, mimicking dmesg output; in addition to "--user"
4154 and "--system" switches for showing only user's own logs
4157 * systemd-delta can now show information about drop-in
4158 snippets extending unit files.
4160 * libsystemd-bus has been substantially updated but is still
4161 not available as public API.
4163 * systemd will now look for the "debug" argument on the kernel
4164 command line and enable debug logging, similar to what
4165 "systemd.log_level=debug" already did before.
4167 * "systemctl set-default", "systemctl get-default" has been
4168 added to configure the default.target symlink, which
4169 controls what to boot into by default.
4171 * "systemctl set-log-level" has been added as a convenient
4172 way to raise and lower systemd logging threshold.
4174 * "systemd-analyze plot" will now show the time the various
4175 generators needed for execution, as well as information
4176 about the unit file loading.
4178 * libsystemd-journal gained a new sd_journal_open_files() call
4179 for opening specific journal files. journactl also gained a
4180 new switch to expose this new functionality. Previously we
4181 only supported opening all files from a directory, or all
4182 files from the system, as opening individual files only is
4183 racy due to journal file rotation.
4185 * systemd gained the new DefaultEnvironment= setting in
4186 /etc/systemd/system.conf to set environment variables for
4189 * If a privileged process logs a journal message with the
4190 OBJECT_PID= field set, then journald will automatically
4191 augment this with additional OBJECT_UID=, OBJECT_GID=,
4192 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4193 system services want to log events about specific client
4194 processes. journactl/systemctl has been updated to make use
4195 of this information if all log messages regarding a specific
4198 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4199 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4200 Reisner, David Coppa, David King, David Strauss, Eelco
4201 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4202 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4203 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4204 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4205 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4206 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4207 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4208 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4209 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4210 Łukasz Stelmach, 장동준
4214 * The Python bindings gained some minimal support for the APIs
4215 exposed by libsystemd-logind.
4217 * ConditionSecurity= gained support for detecting SMACK. Since
4218 this condition already supports SELinux and AppArmor we only
4219 miss IMA for this. Patches welcome!
4221 Contributions from: Karol Lewandowski, Lennart Poettering,
4222 Zbigniew Jędrzejewski-Szmek
4226 * systemd-nspawn will now create /etc/resolv.conf if
4227 necessary, before bind-mounting the host's file onto it.
4229 * systemd-nspawn will now store meta information about a
4230 container on the container's cgroup as extended attribute
4231 fields, including the root directory.
4233 * The cgroup hierarchy has been reworked in many ways. All
4234 objects any of the components systemd creates in the cgroup
4235 tree are now suffixed. More specifically, user sessions are
4236 now placed in cgroups suffixed with ".session", users in
4237 cgroups suffixed with ".user", and nspawn containers in
4238 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4239 names are now escaped in a simple scheme to avoid collision
4240 of userspace object names with kernel filenames. This work
4241 is preparation for making these objects relocatable in the
4242 cgroup tree, in order to allow easy resource partitioning of
4243 these objects without causing naming conflicts.
4245 * systemctl list-dependencies gained the new switches
4246 --plain, --reverse, --after and --before.
4248 * systemd-inhibit now shows the process name of processes that
4249 have taken an inhibitor lock.
4251 * nss-myhostname will now also resolve "localhost"
4252 implicitly. This makes /etc/hosts an optional file and
4253 nicely handles that on IPv6 ::1 maps to both "localhost" and
4256 * libsystemd-logind.so gained a new call
4257 sd_get_machine_names() to enumerate running containers and
4258 VMs (currently only supported by very new libvirt and
4259 nspawn). sd_login_monitor can now be used to watch
4260 VMs/containers coming and going.
4262 * .include is not allowed recursively anymore, and only in
4263 unit files. Usually it is better to use drop-in snippets in
4264 .d/*.conf anyway, as introduced with systemd 198.
4266 * systemd-analyze gained a new "critical-chain" command that
4267 determines the slowest chain of units run during system
4268 boot-up. It is very useful for tracking down where
4269 optimizing boot time is the most beneficial.
4271 * systemd will no longer allow manipulating service paths in
4272 the name=systemd:/system cgroup tree using ControlGroup= in
4273 units. (But is still fine with it in all other dirs.)
4275 * There's a new systemd-nspawn@.service service file that may
4276 be used to easily run nspawn containers as system
4277 services. With the container's root directory in
4278 /var/lib/container/foobar it is now sufficient to run
4279 "systemctl start systemd-nspawn@foobar.service" to boot it.
4281 * systemd-cgls gained a new parameter "--machine" to list only
4282 the processes within a certain container.
4284 * ConditionSecurity= now can check for "apparmor". We still
4285 are lacking checks for SMACK and IMA for this condition
4286 check though. Patches welcome!
4288 * A new configuration file /etc/systemd/sleep.conf has been
4289 added that may be used to configure which kernel operation
4290 systemd is supposed to execute when "suspend", "hibernate"
4291 or "hybrid-sleep" is requested. This makes the new kernel
4292 "freeze" state accessible to the user.
4294 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4295 the passed argument if applicable.
4297 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4298 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4299 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4300 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4301 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4302 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4303 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4308 * The output of 'systemctl list-jobs' got some polishing. The
4309 '--type=' argument may now be passed more than once. A new
4310 command 'systemctl list-sockets' has been added which shows
4311 a list of kernel sockets systemd is listening on with the
4312 socket units they belong to, plus the units these socket
4315 * The experimental libsystemd-bus library got substantial
4316 updates to work in conjunction with the (also experimental)
4317 kdbus kernel project. It works well enough to exchange
4318 messages with some sophistication. Note that kdbus is not
4319 ready yet, and the library is mostly an elaborate test case
4320 for now, and not installable.
4322 * systemd gained a new unit 'systemd-static-nodes.service'
4323 that generates static device nodes earlier during boot, and
4324 can run in conjunction with udev.
4326 * libsystemd-login gained a new call sd_pid_get_user_unit()
4327 to retrieve the user systemd unit a process is running
4328 in. This is useful for systems where systemd is used as
4331 * systemd-nspawn now places all containers in the new /machine
4332 top-level cgroup directory in the name=systemd
4333 hierarchy. libvirt will soon do the same, so that we get a
4334 uniform separation of /system, /user and /machine for system
4335 services, user processes and containers/virtual
4336 machines. This new cgroup hierarchy is also useful to stick
4337 stable names to specific container instances, which can be
4338 recognized later this way (this name may be controlled
4339 via systemd-nspawn's new -M switch). libsystemd-login also
4340 gained a new call sd_pid_get_machine_name() to retrieve the
4341 name of the container/VM a specific process belongs to.
4343 * bootchart can now store its data in the journal.
4345 * libsystemd-journal gained a new call
4346 sd_journal_add_conjunction() for AND expressions to the
4347 matching logic. This can be used to express more complex
4348 logical expressions.
4350 * journactl can now take multiple --unit= and --user-unit=
4353 * The cryptsetup logic now understands the "luks.key=" kernel
4354 command line switch for specifying a file to read the
4355 decryption key from. Also, if a configured key file is not
4356 found the tool will now automatically fall back to prompting
4359 * Python systemd.journal module was updated to wrap recently
4360 added functions from libsystemd-journal. The interface was
4361 changed to bring the low level interface in s.j._Reader
4362 closer to the C API, and the high level interface in
4363 s.j.Reader was updated to wrap and convert all data about
4366 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4367 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4368 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4369 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4370 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4371 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4375 * journalctl --update-catalog now understands a new --root=
4376 option to operate on catalogs found in a different root
4379 * During shutdown after systemd has terminated all running
4380 services a final killing loop kills all remaining left-over
4381 processes. We will now print the name of these processes
4382 when we send SIGKILL to them, since this usually indicates a
4385 * If /etc/crypttab refers to password files stored on
4386 configured mount points automatic dependencies will now be
4387 generated to ensure the specific mount is established first
4388 before the key file is attempted to be read.
4390 * 'systemctl status' will now show information about the
4391 network sockets a socket unit is listening on.
4393 * 'systemctl status' will also shown information about any
4394 drop-in configuration file for units. (Drop-In configuration
4395 files in this context are files such as
4396 /etc/systemd/systemd/foobar.service.d/*.conf)
4398 * systemd-cgtop now optionally shows summed up CPU times of
4399 cgroups. Press '%' while running cgtop to switch between
4400 percentage and absolute mode. This is useful to determine
4401 which cgroups use up the most CPU time over the entire
4402 runtime of the system. systemd-cgtop has also been updated
4403 to be 'pipeable' for processing with further shell tools.
4405 * 'hostnamectl set-hostname' will now allow setting of FQDN
4408 * The formatting and parsing of time span values has been
4409 changed. The parser now understands fractional expressions
4410 such as "5.5h". The formatter will now output fractional
4411 expressions for all time spans under 1min, i.e. "5.123456s"
4412 rather than "5s 123ms 456us". For time spans under 1s
4413 millisecond values are shown, for those under 1ms
4414 microsecond values are shown. This should greatly improve
4415 all time-related output of systemd.
4417 * libsystemd-login and libsystemd-journal gained new
4418 functions for querying the poll() events mask and poll()
4419 timeout value for integration into arbitrary event
4422 * localectl gained the ability to list available X11 keymaps
4423 (models, layouts, variants, options).
4425 * 'systemd-analyze dot' gained the ability to filter for
4426 specific units via shell-style globs, to create smaller,
4427 more useful graphs. I.e. it is now possible to create simple
4428 graphs of all the dependencies between only target units, or
4429 of all units that Avahi has dependencies with.
4431 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4432 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4433 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4434 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4435 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4436 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4437 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4441 * The boot-time readahead implementation for rotating media
4442 will now read the read-ahead data in multiple passes which
4443 consist of all read requests made in equidistant time
4444 intervals. This means instead of strictly reading read-ahead
4445 data in its physical order on disk we now try to find a
4446 middle ground between physical and access time order.
4448 * /etc/os-release files gained a new BUILD_ID= field for usage
4449 on operating systems that provide continuous builds of OS
4452 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4453 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4454 William Douglas, Zbigniew Jędrzejewski-Szmek
4458 * systemd-python gained an API exposing libsystemd-daemon.
4460 * The SMACK setup logic gained support for uploading CIPSO
4463 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4464 ReadOnlyDirectories= and InaccessibleDirectories= has
4465 changed. The private /tmp and /var/tmp directories are now
4466 shared by all processes of a service (which means
4467 ExecStartPre= may now leave data in /tmp that ExecStart= of
4468 the same service can still access). When a service is
4469 stopped its temporary directories are immediately deleted
4470 (normal clean-up with tmpfiles is still done in addition to
4473 * By default, systemd will now set a couple of sysctl
4474 variables in the kernel: the safe sysrq options are turned
4475 on, IP route verification is turned on, and source routing
4476 disabled. The recently added hardlink and softlink
4477 protection of the kernel is turned on. These settings should
4478 be reasonably safe, and good defaults for all new systems.
4480 * The predictable network naming logic may now be turned off
4481 with a new kernel command line switch: net.ifnames=0.
4483 * A new libsystemd-bus module has been added that implements a
4484 pretty complete D-Bus client library. For details see:
4486 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4488 * journald will now explicitly flush the journal files to disk
4489 at the latest 5min after each write. The file will then also
4490 be marked offline until the next write. This should increase
4491 reliability in case of a crash. The synchronization delay
4492 can be configured via SyncIntervalSec= in journald.conf.
4494 * There's a new remote-fs-setup.target unit that can be used
4495 to pull in specific services when at least one remote file
4496 system is to be mounted.
4498 * There are new targets timers.target and paths.target as
4499 canonical targets to pull user timer and path units in
4500 from. This complements sockets.target with a similar
4501 purpose for socket units.
4503 * libudev gained a new call udev_device_set_attribute_value()
4504 to set sysfs attributes of a device.
4506 * The udev daemon now sets the default number of worker
4507 processes executed in parallel based on the number of available
4508 CPUs instead of the amount of available RAM. This is supposed
4509 to provide a more reliable default and limit a too aggressive
4510 paralellism for setups with 1000s of devices connected.
4512 Contributions from: Auke Kok, Colin Walters, Cristian
4513 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4514 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4515 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4516 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4517 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4518 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4519 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4520 Zbigniew Jędrzejewski-Szmek
4524 * Configuration of unit files may now be extended via drop-in
4525 files without having to edit/override the unit files
4526 themselves. More specifically, if the administrator wants to
4527 change one value for a service file foobar.service he can
4528 now do so by dropping in a configuration snippet into
4529 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4530 will load all these snippets and apply them on top of the
4531 main unit configuration file, possibly extending or
4532 overriding its settings. Using these drop-in snippets is
4533 generally nicer than the two earlier options for changing
4534 unit files locally: copying the files from
4535 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4536 them there; or creating a new file in /etc/systemd/system/
4537 that incorporates the original one via ".include". Drop-in
4538 snippets into these .d/ directories can be placed in any
4539 directory systemd looks for units in, and the usual
4540 overriding semantics between /usr/lib, /etc and /run apply
4543 * Most unit file settings which take lists of items can now be
4544 reset by assigning the empty string to them. For example,
4545 normally, settings such as Environment=FOO=BAR append a new
4546 environment variable assignment to the environment block,
4547 each time they are used. By assigning Environment= the empty
4548 string the environment block can be reset to empty. This is
4549 particularly useful with the .d/*.conf drop-in snippets
4550 mentioned above, since this adds the ability to reset list
4551 settings from vendor unit files via these drop-ins.
4553 * systemctl gained a new "list-dependencies" command for
4554 listing the dependencies of a unit recursively.
4556 * Inhibitors are now honored and listed by "systemctl
4557 suspend", "systemctl poweroff" (and similar) too, not only
4558 GNOME. These commands will also list active sessions by
4561 * Resource limits (as exposed by the various control group
4562 controllers) can now be controlled dynamically at runtime
4563 for all units. More specifically, you can now use a command
4564 like "systemctl set-cgroup-attr foobar.service cpu.shares
4565 2000" to alter the CPU shares a specific service gets. These
4566 settings are stored persistently on disk, and thus allow the
4567 administrator to easily adjust the resource usage of
4568 services with a few simple commands. This dynamic resource
4569 management logic is also available to other programs via the
4570 bus. Almost any kernel cgroup attribute and controller is
4573 * systemd-vconsole-setup will now copy all font settings to
4574 all allocated VTs, where it previously applied them only to
4577 * libsystemd-login gained the new sd_session_get_tty() API
4580 * This release drops support for a few legacy or
4581 distribution-specific LSB facility names when parsing init
4582 scripts: $x-display-manager, $mail-transfer-agent,
4583 $mail-transport-agent, $mail-transfer-agent, $smtp,
4584 $null. Also, the mail-transfer-agent.target unit backing
4585 this has been removed. Distributions which want to retain
4586 compatibility with this should carry the burden for
4587 supporting this themselves and patch support for these back
4588 in, if they really need to. Also, the facilities $syslog and
4589 $local_fs are now ignored, since systemd does not support
4590 early-boot LSB init scripts anymore, and these facilities
4591 are implied anyway for normal services. syslog.target has
4594 * There are new bus calls on PID1's Manager object for
4595 cancelling jobs, and removing snapshot units. Previously,
4596 both calls were only available on the Job and Snapshot
4599 * systemd-journal-gatewayd gained SSL support.
4601 * The various "environment" files, such as /etc/locale.conf
4602 now support continuation lines with a backslash ("\") as
4603 last character in the line, similarly in style (but different)
4604 to how this is supported in shells.
4606 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4607 now implicitly appended to every log entry logged. systemctl
4608 has been updated to filter by this field when operating on a
4609 user systemd instance.
4611 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4612 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4613 the container. This makes it easier to boot unmodified
4614 Fedora systems in a container, which however still requires
4615 audit=0 to be passed on the kernel command line. Auditing in
4616 kernel and userspace is unfortunately still too broken in
4617 context of containers, hence we recommend compiling it out
4618 of the kernel or using audit=0. Hopefully this will be fixed
4619 one day for good in the kernel.
4621 * nspawn gained the new --bind= and --bind-ro= parameters to
4622 bind mount specific directories from the host into the
4625 * nspawn will now mount its own devpts file system instance
4626 into the container, in order not to leak pty devices from
4627 the host into the container.
4629 * systemd will now read the firmware boot time performance
4630 information from the EFI variables, if the used boot loader
4631 supports this, and takes it into account for boot performance
4632 analysis via "systemd-analyze". This is currently supported
4633 only in conjunction with Gummiboot, but could be supported
4634 by other boot loaders too. For details see:
4636 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4638 * A new generator has been added that automatically mounts the
4639 EFI System Partition (ESP) to /boot, if that directory
4640 exists, is empty, and no other file system has been
4641 configured to be mounted there.
4643 * logind will now send out PrepareForSleep(false) out
4644 unconditionally, after coming back from suspend. This may be
4645 used by applications as asynchronous notification for
4646 system resume events.
4648 * "systemctl unlock-sessions" has been added, that allows
4649 unlocking the screens of all user sessions at once, similar
4650 to how "systemctl lock-sessions" already locked all users
4651 sessions. This is backed by a new D-Bus call UnlockSessions().
4653 * "loginctl seat-status" will now show the master device of a
4654 seat. (i.e. the device of a seat that needs to be around for
4655 the seat to be considered available, usually the graphics
4658 * tmpfiles gained a new "X" line type, that allows
4659 configuration of files and directories (with wildcards) that
4660 shall be excluded from automatic cleanup ("aging").
4662 * udev default rules set the device node permissions now only
4663 at "add" events, and do not change them any longer with a
4664 later "change" event.
4666 * The log messages for lid events and power/sleep keypresses
4667 now carry a message ID.
4669 * We now have a substantially larger unit test suite, but this
4670 continues to be work in progress.
4672 * udevadm hwdb gained a new --root= parameter to change the
4673 root directory to operate relative to.
4675 * logind will now issue a background sync() request to the kernel
4676 early at shutdown, so that dirty buffers are flushed to disk early
4677 instead of at the last moment, in order to optimize shutdown
4680 * A new bootctl tool has been added that is an interface for
4681 certain boot loader operations. This is currently a preview
4682 and is likely to be extended into a small mechanism daemon
4683 like timedated, localed, hostnamed, and can be used by
4684 graphical UIs to enumerate available boot options, and
4685 request boot into firmware operations.
4687 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4688 the rest of the package. It also has been updated to work
4689 correctly in initrds.
4691 * Policykit previously has been runtime optional, and is now
4692 also compile time optional via a configure switch.
4694 * systemd-analyze has been reimplemented in C. Also "systemctl
4695 dot" has moved into systemd-analyze.
4697 * "systemctl status" with no further parameters will now print
4698 the status of all active or failed units.
4700 * Operations such as "systemctl start" can now be executed
4701 with a new mode "--irreversible" which may be used to queue
4702 operations that cannot accidentally be reversed by a later
4703 job queuing. This is by default used to make shutdown
4704 requests more robust.
4706 * The Python API of systemd now gained a new module for
4707 reading journal files.
4709 * A new tool kernel-install has been added that can install
4710 kernel images according to the Boot Loader Specification:
4712 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4714 * Boot time console output has been improved to provide
4715 animated boot time output for hanging jobs.
4717 * A new tool systemd-activate has been added which can be used
4718 to test socket activation with, directly from the command
4719 line. This should make it much easier to test and debug
4720 socket activation in daemons.
4722 * journalctl gained a new "--reverse" (or -r) option to show
4723 journal output in reverse order (i.e. newest line first).
4725 * journalctl gained a new "--pager-end" (or -e) option to jump
4726 to immediately jump to the end of the journal in the
4727 pager. This is only supported in conjunction with "less".
4729 * journalctl gained a new "--user-unit=" option, that works
4730 similarly to "--unit=" but filters for user units rather than
4733 * A number of unit files to ease adoption of systemd in
4734 initrds has been added. This moves some minimal logic from
4735 the various initrd implementations into systemd proper.
4737 * The journal files are now owned by a new group
4738 "systemd-journal", which exists specifically to allow access
4739 to the journal, and nothing else. Previously, we used the
4740 "adm" group for that, which however possibly covers more
4741 than just journal/log file access. This new group is now
4742 already used by systemd-journal-gatewayd to ensure this
4743 daemon gets access to the journal files and as little else
4744 as possible. Note that "make install" will also set FS ACLs
4745 up for /var/log/journal to give "adm" and "wheel" read
4746 access to it, in addition to "systemd-journal" which owns
4747 the journal files. We recommend that packaging scripts also
4748 add read access to "adm" + "wheel" to /var/log/journal, and
4749 all existing/future journal files. To normal users and
4750 administrators little changes, however packagers need to
4751 ensure to create the "systemd-journal" system group at
4752 package installation time.
4754 * The systemd-journal-gatewayd now runs as unprivileged user
4755 systemd-journal-gateway:systemd-journal-gateway. Packaging
4756 scripts need to create these system user/group at
4759 * timedated now exposes a new boolean property CanNTP that
4760 indicates whether a local NTP service is available or not.
4762 * systemd-detect-virt will now also detect xen PVs
4764 * The pstore file system is now mounted by default, if it is
4767 * In addition to the SELinux and IMA policies we will now also
4768 load SMACK policies at early boot.
4770 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4771 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4772 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4773 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4774 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4775 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4776 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4777 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4778 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4779 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4780 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4781 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4782 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4783 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4787 * Timer units now support calendar time events in addition to
4788 monotonic time events. That means you can now trigger a unit
4789 based on a calendar time specification such as "Thu,Fri
4790 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4791 or fifth day of any month of the year 2013, given that it is
4792 a thursday or friday. This brings timer event support
4793 considerably closer to cron's capabilities. For details on
4794 the supported calendar time specification language see
4797 * udev now supports a number of different naming policies for
4798 network interfaces for predictable names, and a combination
4799 of these policies is now the default. Please see this wiki
4800 document for details:
4802 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4804 * Auke Kok's bootchart implementation has been added to the
4805 systemd tree. It is an optional component that can graph the
4806 boot in quite some detail. It is one of the best bootchart
4807 implementations around and minimal in its code and
4810 * nss-myhostname has been integrated into the systemd source
4811 tree. nss-myhostname guarantees that the local hostname
4812 always stays resolvable via NSS. It has been a weak
4813 requirement of systemd-hostnamed since a long time, and
4814 since its code is actually trivial we decided to just
4815 include it in systemd's source tree. It can be turned off
4816 with a configure switch.
4818 * The read-ahead logic is now capable of properly detecting
4819 whether a btrfs file system is on SSD or rotating media, in
4820 order to optimize the read-ahead scheme. Previously, it was
4821 only capable of detecting this on traditional file systems
4824 * In udev, additional device properties are now read from the
4825 IAB in addition to the OUI database. Also, Bluetooth company
4826 identities are attached to the devices as well.
4828 * In service files %U may be used as specifier that is
4829 replaced by the configured user name of the service.
4831 * nspawn may now be invoked without a controlling TTY. This
4832 makes it suitable for invocation as its own service. This
4833 may be used to set up a simple containerized server system
4834 using only core OS tools.
4836 * systemd and nspawn can now accept socket file descriptors
4837 when they are started for socket activation. This enables
4838 implementation of socket activated nspawn
4839 containers. i.e. think about autospawning an entire OS image
4840 when the first SSH or HTTP connection is received. We expect
4841 that similar functionality will also be added to libvirt-lxc
4844 * journalctl will now suppress ANSI color codes when
4845 presenting log data.
4847 * systemctl will no longer show control group information for
4848 a unit if a the control group is empty anyway.
4850 * logind can now automatically suspend/hibernate/shutdown the
4853 * /etc/machine-info and hostnamed now also expose the chassis
4854 type of the system. This can be used to determine whether
4855 the local system is a laptop, desktop, handset or
4856 tablet. This information may either be configured by the
4857 user/vendor or is automatically determined from ACPI and DMI
4858 information if possible.
4860 * A number of PolicyKit actions are now bound together with
4861 "imply" rules. This should simplify creating UIs because
4862 many actions will now authenticate similar ones as well.
4864 * Unit files learnt a new condition ConditionACPower= which
4865 may be used to conditionalize a unit depending on whether an
4866 AC power source is connected or not, of whether the system
4867 is running on battery power.
4869 * systemctl gained a new "is-failed" verb that may be used in
4870 shell scripts and suchlike to check whether a specific unit
4871 is in the "failed" state.
4873 * The EnvironmentFile= setting in unit files now supports file
4874 globbing, and can hence be used to easily read a number of
4875 environment files at once.
4877 * systemd will no longer detect and recognize specific
4878 distributions. All distribution-specific #ifdeffery has been
4879 removed, systemd is now fully generic and
4880 distribution-agnostic. Effectively, not too much is lost as
4881 a lot of the code is still accessible via explicit configure
4882 switches. However, support for some distribution specific
4883 legacy configuration file formats has been dropped. We
4884 recommend distributions to simply adopt the configuration
4885 files everybody else uses now and convert the old
4886 configuration from packaging scripts. Most distributions
4887 already did that. If that's not possible or desirable,
4888 distributions are welcome to forward port the specific
4889 pieces of code locally from the git history.
4891 * When logging a message about a unit systemd will now always
4892 log the unit name in the message meta data.
4894 * localectl will now also discover system locale data that is
4895 not stored in locale archives, but directly unpacked.
4897 * logind will no longer unconditionally use framebuffer
4898 devices as seat masters, i.e. as devices that are required
4899 to be existing before a seat is considered preset. Instead,
4900 it will now look for all devices that are tagged as
4901 "seat-master" in udev. By default, framebuffer devices will
4902 be marked as such, but depending on local systems, other
4903 devices might be marked as well. This may be used to
4904 integrate graphics cards using closed source drivers (such
4905 as NVidia ones) more nicely into logind. Note however, that
4906 we recommend using the open source NVidia drivers instead,
4907 and no udev rules for the closed-source drivers will be
4908 shipped from us upstream.
4910 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4911 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4912 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4913 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4914 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4915 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4916 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4917 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4918 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4919 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4920 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4921 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4926 * udev gained support for loading additional device properties
4927 from an indexed database that is keyed by vendor/product IDs
4928 and similar device identifiers. For the beginning this
4929 "hwdb" is populated with data from the well-known PCI and
4930 USB database, but also includes PNP, ACPI and OID data. In
4931 the longer run this indexed database shall grow into
4932 becoming the one central database for non-essential
4933 userspace device metadata. Previously, data from the PCI/USB
4934 database was only attached to select devices, since the
4935 lookup was a relatively expensive operation due to O(n) time
4936 complexity (with n being the number of entries in the
4937 database). Since this is now O(1), we decided to add in this
4938 data for all devices where this is available, by
4939 default. Note that the indexed database needs to be rebuilt
4940 when new data files are installed. To achieve this you need
4941 to update your packaging scripts to invoke "udevadm hwdb
4942 --update" after installation of hwdb data files. For
4943 RPM-based distributions we introduced the new
4944 %udev_hwdb_update macro for this purpose.
4946 * The Journal gained support for the "Message Catalog", an
4947 indexed database to link up additional information with
4948 journal entries. For further details please check:
4950 http://www.freedesktop.org/wiki/Software/systemd/catalog
4952 The indexed message catalog database also needs to be
4953 rebuilt after installation of message catalog files. Use
4954 "journalctl --update-catalog" for this. For RPM-based
4955 distributions we introduced the %journal_catalog_update
4956 macro for this purpose.
4958 * The Python Journal bindings gained support for the standard
4959 Python logging framework.
4961 * The Journal API gained new functions for checking whether
4962 the underlying file system of a journal file is capable of
4963 properly reporting file change notifications, or whether
4964 applications that want to reflect journal changes "live"
4965 need to recheck journal files continuously in appropriate
4968 * It is now possible to set the "age" field for tmpfiles
4969 entries to 0, indicating that files matching this entry
4970 shall always be removed when the directories are cleaned up.
4972 * coredumpctl gained a new "gdb" verb which invokes gdb
4973 right-away on the selected coredump.
4975 * There's now support for "hybrid sleep" on kernels that
4976 support this, in addition to "suspend" and "hibernate". Use
4977 "systemctl hybrid-sleep" to make use of this.
4979 * logind's HandleSuspendKey= setting (and related settings)
4980 now gained support for a new "lock" setting to simply
4981 request the screen lock on all local sessions, instead of
4982 actually executing a suspend or hibernation.
4984 * systemd will now mount the EFI variables file system by
4987 * Socket units now gained support for configuration of the
4988 SMACK security label.
4990 * timedatectl will now output the time of the last and next
4991 daylight saving change.
4993 * We dropped support for various legacy and distro-specific
4994 concepts, such as insserv, early-boot SysV services
4995 (i.e. those for non-standard runlevels such as 'b' or 'S')
4996 or ArchLinux /etc/rc.conf support. We recommend the
4997 distributions who still need support this to either continue
4998 to maintain the necessary patches downstream, or find a
4999 different solution. (Talk to us if you have questions!)
5001 * Various systemd components will now bypass PolicyKit checks
5002 for root and otherwise handle properly if PolicyKit is not
5003 found to be around. This should fix most issues for
5004 PolicyKit-less systems. Quite frankly this should have been
5005 this way since day one. It is absolutely our intention to
5006 make systemd work fine on PolicyKit-less systems, and we
5007 consider it a bug if something does not work as it should if
5008 PolicyKit is not around.
5010 * For embedded systems it is now possible to build udev and
5011 systemd without blkid and/or kmod support.
5013 * "systemctl switch-root" is now capable of switching root
5014 more than once. I.e. in addition to transitions from the
5015 initrd to the host OS it is now possible to transition to
5016 further OS images from the host. This is useful to implement
5017 offline updating tools.
5019 * Various other additions have been made to the RPM macros
5020 shipped with systemd. Use %udev_rules_update() after
5021 installing new udev rules files. %_udevhwdbdir,
5022 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5023 %_sysctldir are now available which resolve to the right
5024 directories for packages to place various data files in.
5026 * journalctl gained the new --full switch (in addition to
5027 --all, to disable ellipsation for long messages.
5029 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5030 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5031 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5032 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5033 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5034 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5035 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5036 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5037 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5041 * journalctl gained new --since= and --until= switches to
5042 filter by time. It also now supports nice filtering for
5043 units via --unit=/-u.
5045 * Type=oneshot services may use ExecReload= and do the
5048 * The journal daemon now supports time-based rotation and
5049 vacuuming, in addition to the usual disk-space based
5052 * The journal will now index the available field values for
5053 each field name. This enables clients to show pretty drop
5054 downs of available match values when filtering. The bash
5055 completion of journalctl has been updated
5056 accordingly. journalctl gained a new switch -F to list all
5057 values a certain field takes in the journal database.
5059 * More service events are now written as structured messages
5060 to the journal, and made recognizable via message IDs.
5062 * The timedated, localed and hostnamed mini-services which
5063 previously only provided support for changing time, locale
5064 and hostname settings from graphical DEs such as GNOME now
5065 also have a minimal (but very useful) text-based client
5066 utility each. This is probably the nicest way to changing
5067 these settings from the command line now, especially since
5068 it lists available options and is fully integrated with bash
5071 * There's now a new tool "systemd-coredumpctl" to list and
5072 extract coredumps from the journal.
5074 * We now install a README each in /var/log/ and
5075 /etc/rc.d/init.d explaining where the system logs and init
5076 scripts went. This hopefully should help folks who go to
5077 that dirs and look into the otherwise now empty void and
5078 scratch their heads.
5080 * When user-services are invoked (by systemd --user) the
5081 $MANAGERPID env var is set to the PID of systemd.
5083 * SIGRTMIN+24 when sent to a --user instance will now result
5084 in immediate termination of systemd.
5086 * gatewayd received numerous feature additions such as a
5087 "follow" mode, for live syncing and filtering.
5089 * browse.html now allows filtering and showing detailed
5090 information on specific entries. Keyboard navigation and
5091 mouse screen support has been added.
5093 * gatewayd/journalctl now supports HTML5/JSON
5094 Server-Sent-Events as output.
5096 * The SysV init script compatibility logic will now
5097 heuristically determine whether a script supports the
5098 "reload" verb, and only then make this available as
5101 * "systemctl status --follow" has been removed, use "journalctl
5104 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5105 have been removed since they are hardly useful to be
5108 * And I'd like to take the opportunity to specifically mention
5109 Zbigniew for his great contributions. Zbigniew, you rock!
5111 Contributions from: Andrew Eikum, Christian Hesse, Colin
5112 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5113 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5114 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5115 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5116 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5117 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5121 * If /etc/vconsole.conf is non-existent or empty we will no
5122 longer load any console font or key map at boot by
5123 default. Instead the kernel defaults will be left
5124 intact. This is definitely the right thing to do, as no
5125 configuration should mean no configuration, and hard-coding
5126 font names that are different on all archs is probably a bad
5127 idea. Also, the kernel default key map and font should be
5128 good enough for most cases anyway, and mostly identical to
5129 the userspace fonts/key maps we previously overloaded them
5130 with. If distributions want to continue to default to a
5131 non-kernel font or key map they should ship a default
5132 /etc/vconsole.conf with the appropriate contents.
5134 Contributions from: Colin Walters, Daniel J Walsh, Dave
5135 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5136 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5140 * journalctl gained a new --cursor= switch to show entries
5141 starting from the specified location in the journal.
5143 * We now enforce a size limit on journal entry fields exported
5144 with "-o json" in journalctl. Fields larger than 4K will be
5145 assigned null. This can be turned off with --all.
5147 * An (optional) journal gateway daemon is now available as
5148 "systemd-journal-gatewayd.service". This service provides
5149 access to the journal via HTTP and JSON. This functionality
5150 will be used to implement live log synchronization in both
5151 pull and push modes, but has various other users too, such
5152 as easy log access for debugging of embedded devices. Right
5153 now it is already useful to retrieve the journal via HTTP:
5155 # systemctl start systemd-journal-gatewayd.service
5156 # wget http://localhost:19531/entries
5158 This will download the journal contents in a
5159 /var/log/messages compatible format. The same as JSON:
5161 # curl -H"Accept: application/json" http://localhost:19531/entries
5163 This service is also accessible via a web browser where a
5164 single static HTML5 app is served that uses the JSON logic
5165 to enable the user to do some basic browsing of the
5166 journal. This will be extended later on. Here's an example
5167 screenshot of this app in its current state:
5169 http://0pointer.de/public/journal-gatewayd
5171 Contributions from: Kay Sievers, Lennart Poettering, Robert
5172 Milasan, Tom Gundersen
5176 * The bash completion logic is now available for journalctl
5179 * We do not mount the "cpuset" controller anymore together with
5180 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5181 started if no parameters are assigned to it. "cpuset" hence
5182 broke code that assumed it it could create "cpu" groups and
5185 * journalctl -f will now subscribe to terminal size changes,
5186 and line break accordingly.
5188 Contributions from: Dave Reisner, Kay Sievers, Lennart
5189 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5193 * nspawn will now create a symlink /etc/localtime in the
5194 container environment, copying the host's timezone
5195 setting. Previously this has been done via a bind mount, but
5196 since symlinks cannot be bind mounted this has now been
5197 changed to create/update the appropriate symlink.
5199 * journalctl -n's line number argument is now optional, and
5200 will default to 10 if omitted.
5202 * journald will now log the maximum size the journal files may
5203 take up on disk. This is particularly useful if the default
5204 built-in logic of determining this parameter from the file
5205 system size is used. Use "systemctl status
5206 systemd-journald.service" to see this information.
5208 * The multi-seat X wrapper tool has been stripped down. As X
5209 is now capable of enumerating graphics devices via udev in a
5210 seat-aware way the wrapper is not strictly necessary
5211 anymore. A stripped down temporary stop-gap is still shipped
5212 until the upstream display managers have been updated to
5213 fully support the new X logic. Expect this wrapper to be
5214 removed entirely in one of the next releases.
5216 * HandleSleepKey= in logind.conf has been split up into
5217 HandleSuspendKey= and HandleHibernateKey=. The old setting
5218 is not available anymore. X11 and the kernel are
5219 distinguishing between these keys and we should too. This
5220 also means the inhibition lock for these keys has been split
5223 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5224 Poettering, Lukas Nykryn, Václav Pavlín
5228 * Whenever a unit changes state we will now log this to the
5229 journal and show along the unit's own log output in
5232 * ConditionPathIsMountPoint= can now properly detect bind
5233 mount points too. (Previously, a bind mount of one file
5234 system to another place in the same file system could not be
5235 detected as mount, since they shared struct stat's st_dev
5238 * We will now mount the cgroup controllers cpu, cpuacct,
5239 cpuset and the controllers net_cls, net_prio together by
5242 * nspawn containers will now have a virtualized boot
5243 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5244 over with a randomized ID at container initialization). This
5245 has the effect of making "journalctl -b" do the right thing
5248 * The JSON output journal serialization has been updated not
5249 to generate "endless" list objects anymore, but rather one
5250 JSON object per line. This is more in line how most JSON
5251 parsers expect JSON objects. The new output mode
5252 "json-pretty" has been added to provide similar output, but
5253 neatly aligned for readability by humans.
5255 * We dropped all explicit sync() invocations in the shutdown
5256 code. The kernel does this implicitly anyway in the kernel
5257 reboot() syscall. halt(8)'s -n option is now a compatibility
5260 * We now support virtualized reboot() in containers, as
5261 supported by newer kernels. We will fall back to exit() if
5262 CAP_SYS_REBOOT is not available to the container. Also,
5263 nspawn makes use of this now and will actually reboot the
5264 container if the containerized OS asks for that.
5266 * journalctl will only show local log output by default
5267 now. Use --merge (-m) to show remote log output, too.
5269 * libsystemd-journal gained the new sd_journal_get_usage()
5270 call to determine the current disk usage of all journal
5271 files. This is exposed in the new "journalctl --disk-usage"
5274 * journald gained a new configuration setting SplitMode= in
5275 journald.conf which may be used to control how user journals
5276 are split off. See journald.conf(5) for details.
5278 * A new condition type ConditionFileNotEmpty= has been added.
5280 * tmpfiles' "w" lines now support file globbing, to write
5281 multiple files at once.
5283 * We added Python bindings for the journal submission
5284 APIs. More Python APIs for a number of selected APIs will
5285 likely follow. Note that we intend to add native bindings
5286 only for the Python language, as we consider it common
5287 enough to deserve bindings shipped within systemd. There are
5288 various projects outside of systemd that provide bindings
5289 for languages such as PHP or Lua.
5291 * Many conditions will now resolve specifiers such as %i. In
5292 addition, PathChanged= and related directives of .path units
5293 now support specifiers as well.
5295 * There's now a new RPM macro definition for the system preset
5298 * journald will now warn if it ca not forward a message to the
5299 syslog daemon because its socket is full.
5301 * timedated will no longer write or process /etc/timezone,
5302 except on Debian. As we do not support late mounted /usr
5303 anymore /etc/localtime always being a symlink is now safe,
5304 and hence the information in /etc/timezone is not necessary
5307 * logind will now always reserve one VT for a text getty (VT6
5308 by default). Previously if more than 6 X sessions where
5309 started they took up all the VTs with auto-spawned gettys,
5310 so that no text gettys were available anymore.
5312 * udev will now automatically inform the btrfs kernel logic
5313 about btrfs RAID components showing up. This should make
5314 simple hotplug based btrfs RAID assembly work.
5316 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5317 (but not for its children which will stay at the kernel
5318 default). This should allow setups with a lot more listening
5321 * systemd will now always pass the configured timezone to the
5322 kernel at boot. timedated will do the same when the timezone
5325 * logind's inhibition logic has been updated. By default,
5326 logind will now handle the lid switch, the power and sleep
5327 keys all the time, even in graphical sessions. If DEs want
5328 to handle these events on their own they should take the new
5329 handle-power-key, handle-sleep-key and handle-lid-switch
5330 inhibitors during their runtime. A simple way to achieve
5331 that is to invoke the DE wrapped in an invocation of:
5333 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5335 * Access to unit operations is now checked via SELinux taking
5336 the unit file label and client process label into account.
5338 * systemd will now notify the administrator in the journal
5339 when he over-mounts a non-empty directory.
5341 * There are new specifiers that are resolved in unit files,
5342 for the host name (%H), the machine ID (%m) and the boot ID
5345 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5346 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5347 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5348 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5349 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5350 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5351 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5355 * Support for reading structured kernel messages from
5356 /dev/kmsg has now been added and is enabled by default.
5358 * Support for reading kernel messages from /proc/kmsg has now
5359 been removed. If you want kernel messages in the journal
5360 make sure to run a recent kernel (>= 3.5) that supports
5361 reading structured messages from /dev/kmsg (see
5362 above). /proc/kmsg is now exclusive property of classic
5363 syslog daemons again.
5365 * The libudev API gained the new
5366 udev_device_new_from_device_id() call.
5368 * The logic for file system namespace (ReadOnlyDirectory=,
5369 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5370 require pivot_root() anymore. This means fewer temporary
5371 directories are created below /tmp for this feature.
5373 * nspawn containers will now see and receive all submounts
5374 made on the host OS below the root file system of the
5377 * Forward Secure Sealing is now supported for Journal files,
5378 which provide cryptographical sealing of journal files so
5379 that attackers cannot alter log history anymore without this
5380 being detectable. Lennart will soon post a blog story about
5381 this explaining it in more detail.
5383 * There are two new service settings RestartPreventExitStatus=
5384 and SuccessExitStatus= which allow configuration of exit
5385 status (exit code or signal) which will be excepted from the
5386 restart logic, resp. consider successful.
5388 * journalctl gained the new --verify switch that can be used
5389 to check the integrity of the structure of journal files and
5390 (if Forward Secure Sealing is enabled) the contents of
5393 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5394 and similar symlinks pre-created. This makes running shells
5395 as container init process a lot more fun.
5397 * The fstab support can now handle PARTUUID= and PARTLABEL=
5400 * A new ConditionHost= condition has been added to match
5401 against the hostname (with globs) and machine ID. This is
5402 useful for clusters where a single OS image is used to
5403 provision a large number of hosts which shall run slightly
5404 different sets of services.
5406 * Services which hit the restart limit will now be placed in a
5409 Contributions from: Bertram Poettering, Dave Reisner, Huang
5410 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5411 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5415 * When running in --user mode systemd will now become a
5416 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5417 tree a lot more organized.
5419 * A new PartOf= unit dependency type has been introduced that
5420 may be used to group services in a natural way.
5422 * "systemctl enable" may now be used to enable instances of
5425 * journalctl now prints error log levels in red, and
5426 warning/notice log levels in bright white. It also supports
5427 filtering by log level now.
5429 * cgtop gained a new -n switch (similar to top), to configure
5430 the maximum number of iterations to run for. It also gained
5431 -b, to run in batch mode (accepting no input).
5433 * The suffix ".service" may now be omitted on most systemctl
5434 command lines involving service unit names.
5436 * There's a new bus call in logind to lock all sessions, as
5437 well as a loginctl verb for it "lock-sessions".
5439 * libsystemd-logind.so gained a new call sd_journal_perror()
5440 that works similar to libc perror() but logs to the journal
5441 and encodes structured information about the error number.
5443 * /etc/crypttab entries now understand the new keyfile-size=
5446 * shutdown(8) now can send a (configurable) wall message when
5447 a shutdown is cancelled.
5449 * The mount propagation mode for the root file system will now
5450 default to "shared", which is useful to make containers work
5451 nicely out-of-the-box so that they receive new mounts from
5452 the host. This can be undone locally by running "mount
5453 --make-rprivate /" if needed.
5455 * The prefdm.service file has been removed. Distributions
5456 should maintain this unit downstream if they intend to keep
5457 it around. However, we recommend writing normal unit files
5458 for display managers instead.
5460 * Since systemd is a crucial part of the OS we will now
5461 default to a number of compiler switches that improve
5462 security (hardening) such as read-only relocations, stack
5463 protection, and suchlike.
5465 * The TimeoutSec= setting for services is now split into
5466 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5467 of individual time outs for the start and the stop phase of
5470 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5471 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5472 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5473 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5474 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5475 Gundersen, Zbigniew Jędrzejewski-Szmek
5479 * The journal and id128 C APIs are now fully documented as man
5482 * Extra safety checks have been added when transitioning from
5483 the initial RAM disk to the main system to avoid accidental
5486 * /etc/crypttab entries now understand the new keyfile-offset=
5489 * systemctl -t can now be used to filter by unit load state.
5491 * The journal C API gained the new sd_journal_wait() call to
5492 make writing synchronous journal clients easier.
5494 * journalctl gained the new -D switch to show journals from a
5497 * journalctl now displays a special marker between log
5498 messages of two different boots.
5500 * The journal is now explicitly flushed to /var via a service
5501 systemd-journal-flush.service, rather than implicitly simply
5502 by seeing /var/log/journal to be writable.
5504 * journalctl (and the journal C APIs) can now match for much
5505 more complex expressions, with alternatives and
5508 * When transitioning from the initial RAM disk to the main
5509 system we will now kill all processes in a killing spree to
5510 ensure no processes stay around by accident.
5512 * Three new specifiers may be used in unit files: %u, %h, %s
5513 resolve to the user name, user home directory resp. user
5514 shell. This is useful for running systemd user instances.
5516 * We now automatically rotate journal files if their data
5517 object hash table gets a fill level > 75%. We also size the
5518 hash table based on the configured maximum file size. This
5519 together should lower hash collisions drastically and thus
5520 speed things up a bit.
5522 * journalctl gained the new "--header" switch to introspect
5523 header data of journal files.
5525 * A new setting SystemCallFilters= has been added to services
5526 which may be used to apply blacklists or whitelists to
5527 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5529 * nspawn gained a new --link-journal= switch (and quicker: -j)
5530 to link the container journal with the host. This makes it
5531 very easy to centralize log viewing on the host for all
5532 guests while still keeping the journal files separated.
5534 * Many bugfixes and optimizations
5536 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5537 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5538 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5543 * Several tools now understand kernel command line arguments,
5544 which are only read when run in an initial RAM disk. They
5545 usually follow closely their normal counterparts, but are
5548 * There's a new tool to analyze the readahead files that are
5549 automatically generated at boot. Use:
5551 /usr/lib/systemd/systemd-readahead analyze /.readahead
5553 * We now provide an early debug shell on tty9 if this enabled. Use:
5555 systemctl enable debug-shell.service
5557 * All plymouth related units have been moved into the Plymouth
5558 package. Please make sure to upgrade your Plymouth version
5561 * systemd-tmpfiles now supports getting passed the basename of
5562 a configuration file only, in which case it will look for it
5563 in all appropriate directories automatically.
5565 * udevadm info now takes a /dev or /sys path as argument, and
5566 does the right thing. Example:
5568 udevadm info /dev/sda
5569 udevadm info /sys/class/block/sda
5571 * systemctl now prints a warning if a unit is stopped but a
5572 unit that might trigger it continues to run. Example: a
5573 service is stopped but the socket that activates it is left
5576 * "systemctl status" will now mention if the log output was
5577 shortened due to rotation since a service has been started.
5579 * The journal API now exposes functions to determine the
5580 "cutoff" times due to rotation.
5582 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5583 immediately flushing of runtime logs to /var if possible,
5584 resp. for triggering immediate rotation of the journal
5587 * It is now considered an error if a service is attempted to
5588 be stopped that is not loaded.
5590 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5592 * systemd-analyze now supports Python 3
5594 * tmpfiles now supports cleaning up directories via aging
5595 where the first level dirs are always kept around but
5596 directories beneath it automatically aged. This is enabled
5597 by prefixing the age field with '~'.
5599 * Seat objects now expose CanGraphical, CanTTY properties
5600 which is required to deal with very fast bootups where the
5601 display manager might be running before the graphics drivers
5602 completed initialization.
5604 * Seat objects now expose a State property.
5606 * We now include RPM macros for service enabling/disabling
5607 based on the preset logic. We recommend RPM based
5608 distributions to make use of these macros if possible. This
5609 makes it simpler to reuse RPM spec files across
5612 * We now make sure that the collected systemd unit name is
5613 always valid when services log to the journal via
5616 * There's a new man page kernel-command-line(7) detailing all
5617 command line options we understand.
5619 * The fstab generator may now be disabled at boot by passing
5620 fstab=0 on the kernel command line.
5622 * A new kernel command line option modules-load= is now understood
5623 to load a specific kernel module statically, early at boot.
5625 * Unit names specified on the systemctl command line are now
5626 automatically escaped as needed. Also, if file system or
5627 device paths are specified they are automatically turned
5628 into the appropriate mount or device unit names. Example:
5630 systemctl status /home
5631 systemctl status /dev/sda
5633 * The SysVConsole= configuration option has been removed from
5634 system.conf parsing.
5636 * The SysV search path is no longer exported on the D-Bus
5639 * The Names= option is been removed from unit file parsing.
5641 * There's a new man page bootup(7) detailing the boot process.
5643 * Every unit and every generator we ship with systemd now
5644 comes with full documentation. The self-explanatory boot is
5647 * A couple of services gained "systemd-" prefixes in their
5648 name if they wrap systemd code, rather than only external
5649 code. Among them fsck@.service which is now
5650 systemd-fsck@.service.
5652 * The HaveWatchdog property has been removed from the D-Bus
5655 * systemd.confirm_spawn= on the kernel command line should now
5658 * There's a new man page crypttab(5) which details all options
5659 we actually understand.
5661 * systemd-nspawn gained a new --capability= switch to pass
5662 additional capabilities to the container.
5664 * timedated will now read known NTP implementation unit names
5665 from /usr/lib/systemd/ntp-units.d/*.list,
5666 systemd-timedated-ntp.target has been removed.
5668 * journalctl gained a new switch "-b" that lists log data of
5669 the current boot only.
5671 * The notify socket is in the abstract namespace again, in
5672 order to support daemons which chroot() at start-up.
5674 * There is a new Storage= configuration option for journald
5675 which allows configuration of where log data should go. This
5676 also provides a way to disable journal logging entirely, so
5677 that data collected is only forwarded to the console, the
5678 kernel log buffer or another syslog implementation.
5680 * Many bugfixes and optimizations
5682 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5683 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5684 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5685 Shawn Landden, Tom Gundersen
5689 * "systemctl help <unit>" now shows the man page if one is
5692 * Several new man pages have been added.
5694 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5695 MaxLevelConsole= can now be specified in
5696 journald.conf. These options allow reducing the amount of
5697 data stored on disk or forwarded by the log level.
5699 * TimerSlackNSec= can now be specified in system.conf for
5700 PID1. This allows system-wide power savings.
5702 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5703 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5708 * logind is now capable of (optionally) handling power and
5709 sleep keys as well as the lid switch.
5711 * journalctl now understands the syntax "journalctl
5712 /usr/bin/avahi-daemon" to get all log output of a specific
5715 * CapabilityBoundingSet= in system.conf now also influences
5716 the capability bound set of usermode helpers of the kernel.
5718 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5719 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5720 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5721 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5725 * Note that we skipped 139 releases here in order to set the
5726 new version to something that is greater than both udev's
5727 and systemd's most recent version number.
5729 * udev: all udev sources are merged into the systemd source tree now.
5730 All future udev development will happen in the systemd tree. It
5731 is still fully supported to use the udev daemon and tools without
5732 systemd running, like in initramfs or other init systems. Building
5733 udev though, will require the *build* of the systemd tree, but
5734 udev can be properly *run* without systemd.
5736 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5737 should be used to create dead device nodes as workarounds for broken
5740 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5741 no longer supported. udev_monitor_new_from_netlink() needs to be
5742 used to subscribe to events.
5744 * udev: when udevd is started by systemd, processes which are left
5745 behind by forking them off of udev rules, are unconditionally cleaned
5746 up and killed now after the event handling has finished. Services or
5747 daemons must be started as systemd services. Services can be
5748 pulled-in by udev to get started, but they can no longer be directly
5749 forked by udev rules.
5751 * udev: the daemon binary is called systemd-udevd now and installed
5752 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5753 to adapt to that, create symlink, or rename the binary after building
5756 * libudev no longer provides these symbols:
5757 udev_monitor_from_socket()
5758 udev_queue_get_failed_list_entry()
5759 udev_get_{dev,sys,run}_path()
5760 The versions number was bumped and symbol versioning introduced.
5762 * systemd-loginctl and systemd-journalctl have been renamed
5763 to loginctl and journalctl to match systemctl.
5765 * The config files: /etc/systemd/systemd-logind.conf and
5766 /etc/systemd/systemd-journald.conf have been renamed to
5767 logind.conf and journald.conf. Package updates should rename
5768 the files to the new names on upgrade.
5770 * For almost all files the license is now LGPL2.1+, changed
5771 from the previous GPL2.0+. Exceptions are some minor stuff
5772 of udev (which will be changed to LGPL2.1 eventually, too),
5773 and the MIT licensed sd-daemon.[ch] library that is suitable
5774 to be used as drop-in files.
5776 * systemd and logind now handle system sleep states, in
5777 particular suspending and hibernating.
5779 * logind now implements a sleep/shutdown/idle inhibiting logic
5780 suitable for a variety of uses. Soonishly Lennart will blog
5781 about this in more detail.
5783 * var-run.mount and var-lock.mount are no longer provided
5784 (which prevously bind mounted these directories to their new
5785 places). Distributions which have not converted these
5786 directories to symlinks should consider stealing these files
5787 from git history and add them downstream.
5789 * We introduced the Documentation= field for units and added
5790 this to all our shipped units. This is useful to make it
5791 easier to explore the boot and the purpose of the various
5794 * All smaller setup units (such as
5795 systemd-vconsole-setup.service) now detect properly if they
5796 are run in a container and are skipped when
5797 appropriate. This guarantees an entirely noise-free boot in
5798 Linux container environments such as systemd-nspawn.
5800 * A framework for implementing offline system updates is now
5801 integrated, for details see:
5802 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5804 * A new service type Type=idle is available now which helps us
5805 avoiding ugly interleaving of getty output and boot status
5808 * There's now a system-wide CapabilityBoundingSet= option to
5809 globally reduce the set of capabilities for the
5810 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5811 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5812 even CAP_NET_ADMIN system-wide for secure systems.
5814 * There are now system-wide DefaultLimitXXX= options to
5815 globally change the defaults of the various resource limits
5816 for all units started by PID 1.
5818 * Harald Hoyer's systemd test suite has been integrated into
5819 systemd which allows easy testing of systemd builds in qemu
5820 and nspawn. (This is really awesome! Ask us for details!)
5822 * The fstab parser is now implemented as generator, not inside
5825 * systemctl will now warn you if .mount units generated from
5826 /etc/fstab are out of date due to changes in fstab that
5827 have not been read by systemd yet.
5829 * systemd is now suitable for usage in initrds. Dracut has
5830 already been updated to make use of this. With this in place
5831 initrds get a slight bit faster but primarily are much
5832 easier to introspect and debug since "systemctl status" in
5833 the host system can be used to introspect initrd services,
5834 and the journal from the initrd is kept around too.
5836 * systemd-delta has been added, a tool to explore differences
5837 between user/admin configuration and vendor defaults.
5839 * PrivateTmp= now affects both /tmp and /var/tmp.
5841 * Boot time status messages are now much prettier and feature
5842 proper english language. Booting up systemd has never been
5845 * Read-ahead pack files now include the inode number of all
5846 files to pre-cache. When the inode changes the pre-caching
5847 is not attempted. This should be nicer to deal with updated
5848 packages which might result in changes of read-ahead
5851 * We now temporaritly lower the kernel's read_ahead_kb variable
5852 when collecting read-ahead data to ensure the kernel's
5853 built-in read-ahead does not add noise to our measurements
5854 of necessary blocks to pre-cache.
5856 * There's now RequiresMountsFor= to add automatic dependencies
5857 for all mounts necessary for a specific file system path.
5859 * MountAuto= and SwapAuto= have been removed from
5860 system.conf. Mounting file systems at boot has to take place
5863 * nspawn now learned a new switch --uuid= to set the machine
5864 ID on the command line.
5866 * nspawn now learned the -b switch to automatically search
5869 * vt102 is now the default TERM for serial TTYs, upgraded from
5872 * systemd-logind now works on VT-less systems.
5874 * The build tree has been reorganized. The individual
5875 components now have directories of their own.
5877 * A new condition type ConditionPathIsReadWrite= is now available.
5879 * nspawn learned the new -C switch to create cgroups for the
5880 container in other hierarchies.
5882 * We now have support for hardware watchdogs, configurable in
5885 * The scheduled shutdown logic now has a public API.
5887 * We now mount /tmp as tmpfs by default, but this can be
5888 masked and /etc/fstab can override it.
5890 * Since udisks does not make use of /media anymore we are not
5891 mounting a tmpfs on it anymore.
5893 * journalctl gained a new --local switch to only interleave
5894 locally generated journal files.
5896 * We can now load the IMA policy at boot automatically.
5898 * The GTK tools have been split off into a systemd-ui.
5900 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5901 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5902 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5903 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5904 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5905 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5906 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5907 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5908 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5913 * This is mostly a bugfix release
5915 * Support optional initialization of the machine ID from the
5916 KVM or container configured UUID.
5918 * Support immediate reboots with "systemctl reboot -ff"
5920 * Show /etc/os-release data in systemd-analyze output
5922 * Many bugfixes for the journal, including endianness fixes and
5923 ensuring that disk space enforcement works
5925 * sd-login.h is C++ comptaible again
5927 * Extend the /etc/os-release format on request of the Debian
5930 * We now refuse non-UTF8 strings used in various configuration
5931 and unit files. This is done to ensure we do not pass invalid
5932 data over D-Bus or expose it elsewhere.
5934 * Register Mimo USB Screens as suitable for automatic seat
5937 * Read SELinux client context from journal clients in a race
5940 * Reorder configuration file lookup order. /etc now always
5941 overrides /run in order to allow the administrator to always
5942 and unconditionally override vendor-supplied or
5943 automatically generated data.
5945 * The various user visible bits of the journal now have man
5946 pages. We still lack man pages for the journal API calls
5949 * We now ship all man pages in HTML format again in the
5952 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5953 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5954 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5959 * This is mostly a bugfix release
5961 * systems lacking /etc/os-release are no longer supported.
5963 * Various functionality updates to libsystemd-login.so
5965 * Track class of PAM logins to distinguish greeters from
5968 Contributions from: Kay Sievers, Lennart Poettering, Michael
5973 * This is an important bugfix release for v41.
5975 * Building man pages is now optional which should be useful
5976 for those building systemd from git but unwilling to install
5979 * Watchdog support for supervising services is now usable. In
5980 a future release support for hardware watchdogs
5981 (i.e. /dev/watchdog) will be added building on this.
5983 * Service start rate limiting is now configurable and can be
5984 turned off per service. When a start rate limit is hit a
5985 reboot can automatically be triggered.
5987 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5989 Contributions from: Benjamin Franzke, Bill Nottingham,
5990 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5991 Schmidt, Michał Górny, Piotr Drąg
5995 * The systemd binary is installed /usr/lib/systemd/systemd now;
5996 An existing /sbin/init symlink needs to be adapted with the
5999 * The code that loads kernel modules has been ported to invoke
6000 libkmod directly, instead of modprobe. This means we do not
6001 support systems with module-init-tools anymore.
6003 * Watchdog support is now already useful, but still not
6006 * A new kernel command line option systemd.setenv= is
6007 understood to set system wide environment variables
6008 dynamically at boot.
6010 * We now limit the set of capabilities of systemd-journald.
6012 * We now set SIGPIPE to ignore by default, since it only is
6013 useful in shell pipelines, and has little use in general
6014 code. This can be disabled with IgnoreSIPIPE=no in unit
6017 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6018 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6023 * This is mostly a bugfix release
6025 * We now expose the reason why a service failed in the
6026 "Result" D-Bus property.
6028 * Rudimentary service watchdog support (will be completed over
6029 the next few releases.)
6031 * When systemd forks off in order execute some service we will
6032 now immediately changes its argv[0] to reflect which process
6033 it will execute. This is useful to minimize the time window
6034 with a generic argv[0], which makes bootcharts more useful
6036 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6037 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6038 Mike Kazantsev, Ray Strode
6042 * This is mostly a test release, but incorporates many
6045 * New systemd-cgtop tool to show control groups by their
6048 * Linking against libacl for ACLs is optional again. If
6049 disabled, support tracking device access for active logins
6050 goes becomes unavailable, and so does access to the user
6051 journals by the respective users.
6053 * If a group "adm" exists, journal files are automatically
6054 owned by them, thus allow members of this group full access
6055 to the system journal as well as all user journals.
6057 * The journal now stores the SELinux context of the logging
6058 client for all entries.
6060 * Add C++ inclusion guards to all public headers
6062 * New output mode "cat" in the journal to print only text
6063 messages, without any meta data like date or time.
6065 * Include tiny X server wrapper as a temporary stop-gap to
6066 teach XOrg udev display enumeration. This is used by display
6067 managers such as gdm, and will go away as soon as XOrg
6068 learned native udev hotplugging for display devices.
6070 * Add new systemd-cat tool for executing arbitrary programs
6071 with STDERR/STDOUT connected to the journal. Can also act as
6072 BSD logger replacement, and does so by default.
6074 * Optionally store all locally generated coredumps in the
6075 journal along with meta data.
6077 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6078 writing short strings to files (for usage for /sys), and for
6079 creating symlinks, character and block device nodes.
6081 * New unit file option ControlGroupPersistent= to make cgroups
6082 persistent, following the mechanisms outlined in
6083 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6085 * Support multiple local RTCs in a sane way
6087 * No longer monopolize IO when replaying readahead data on
6088 rotating disks, since we might starve non-file-system IO to
6089 death, since fanotify() will not see accesses done by blkid,
6092 * Do not show kernel threads in systemd-cgls anymore, unless
6093 requested with new -k switch.
6095 Contributions from: Dan Horák, Kay Sievers, Lennart
6096 Poettering, Michal Schmidt
6100 * This is mostly a test release, but incorporates many
6103 * The git repository moved to:
6104 git://anongit.freedesktop.org/systemd/systemd
6105 ssh://git.freedesktop.org/git/systemd/systemd
6107 * First release with the journal
6108 http://0pointer.de/blog/projects/the-journal.html
6110 * The journal replaces both systemd-kmsg-syslogd and
6111 systemd-stdout-bridge.
6113 * New sd_pid_get_unit() API call in libsystemd-logind
6115 * Many systemadm clean-ups
6117 * Introduce remote-fs-pre.target which is ordered before all
6118 remote mounts and may be used to start services before all
6121 * Added Mageia support
6123 * Add bash completion for systemd-loginctl
6125 * Actively monitor PID file creation for daemons which exit in
6126 the parent process before having finished writing the PID
6127 file in the daemon process. Daemons which do this need to be
6128 fixed (i.e. PID file creation must have finished before the
6129 parent exits), but we now react a bit more gracefully to them.
6131 * Add colourful boot output, mimicking the well-known output
6132 of existing distributions.
6134 * New option PassCredentials= for socket units, for
6135 compatibility with a recent kernel ABI breakage.
6137 * /etc/rc.local is now hooked in via a generator binary, and
6138 thus will no longer act as synchronization point during
6141 * systemctl list-unit-files now supports --root=.
6143 * systemd-tmpfiles now understands two new commands: z, Z for
6144 relabelling files according to the SELinux database. This is
6145 useful to apply SELinux labels to specific files in /sys,
6148 * Output of SysV services is now forwarded to both the console
6149 and the journal by default, not only just the console.
6151 * New man pages for all APIs from libsystemd-login.
6153 * The build tree got reorganized and a the build system is a
6154 lot more modular allowing embedded setups to specifically
6155 select the components of systemd they are interested in.
6157 * Support for Linux systems lacking the kernel VT subsystem is
6160 * configure's --with-rootdir= got renamed to
6161 --with-rootprefix= to follow the naming used by udev and
6164 * Unless specified otherwise we will now install to /usr instead
6165 of /usr/local by default.
6167 * Processes with '@' in argv[0][0] are now excluded from the
6168 final shut-down killing spree, following the logic explained
6170 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6172 * All processes remaining in a service cgroup when we enter
6173 the START or START_PRE states are now killed with
6174 SIGKILL. That means it is no longer possible to spawn
6175 background processes from ExecStart= lines (which was never
6176 supported anyway, and bad style).
6178 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6179 reloading of units together.
6181 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6182 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6183 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6184 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6185 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob